The Germinal centre-derived lymphomas seen through their cellular microenvironment

Authors


Antonino Carbone, MD, Chairman Department of Pathology, Istituto Nazionale Tumori, Milano, Via Venezian 1, 20133 Milan, Italy. E-mail: antonino.carbone@istitutotumori.mi.it

Summary

The human lymph node is a complex tissue resulting from the microenvironmental organisation of different cell populations linked by topographical and/or functional relationships. Germinal centres (GCs) of lymphoid follicles contain a meshwork of follicular dendritic cells in addition to B-cells and some CD4+ T cells. Moreover, there is a sharp demarcation around the whole follicle centre, which is highlighted by fibroblastic reticulum cells. On the whole, GC exerts a role in B cell physiology and malignancy. In GC-derived lymphomas, gene expression profiling studies have raised the possibility that survival of the affected patients may be associated with signatures preferentially expressed in non-malignant T cells and macrophages and/or dendritic cells. Immunohistological analyses in lymphoma biopsy samples have confirmed that the biological behaviour and tumour progression may be influenced by the tumour microenvironment. This review will examine GC-derived lymphomas, including follicular lymphomas, Hodgkin lymphomas and angioimmunoblastic T-cell lymphoma, through their integrated cellular microenvironment, highlighting those findings which may serve as a useful surrogate marker for tumour diagnosis or tumour progression, together with key molecules involved in tumour development.

The human lymph node is a complex tissue resulting from the microenvironmental organisation of different cell populations (lymphoid cells, accessory or non-lymphoid cells, and stromal cells) linked by topographical and/or functional relationships. The primary follicle is a structure made of recirculating immunoglobulin (Ig)M+IgD+ B cells within a network of follicular dendritic cells (FDCs). Here, B cells start to proliferate rapidly and push the IgM+IgD+ B cells aside to form the mantle zone around the germinal centre (GC), yielding a structure known as the secondary follicle (Thorbecke et al, 1994; Han et al, 1997; Natkunam, 2007). After a few days of vigorous proliferation, the characteristic structure of the GC becomes apparent: a dark zone consisting almost exclusively of densely packed proliferating B cells known as centroblasts, and a light zone comprised of smaller, non-dividing centrocytes situated within a mesh of FDCs, T cells and macrophages (Hollowood & Goodlad, 1998; Natkunam, 2007).

On the whole, the GC exerts a role in B-cell physiology and malignancy. Recent studies on normal and malignant B cells have provided new insights into the specialized physiology of the GC response and in the identification of the mechanisms that guide the trafficking of B cells (for a review, see Klein & Dalla-Favera, 2008).

Germinal centres contain a few different microenvironmental zones that have been clearly described. The ‘dark’ zone is considered to be at the base, and the ‘light’ zone is at the apex of the follicular centre. The dark zone contains tightly packed centroblasts, and has by far the highest mitotic activity, whereas the light zone is much less dense. The FDC network extends throughout the GC, but is much sparser in the dark than in the light zone. Some CD4+ T cells are found in the light zones (discussed below). There is a sharp demarcation around the whole follicle centre, which is highlighted by fibroblastic reticulum cells (FRCs) (Gloghini et al, 1990). Tingible body macrophages (TBMs), containing nuclear debris in their cytoplasm from B cells that have recently been in cell cycle, are located throughout the GCs, but their number is highest in the basal light zone (Gloghini & Carbone, 1993; MacLennan, 1994) (Fig 1A).

Figure 1.

 The follicle and its germinal centre (GC). (A) The GC in lymphoid organs is a dynamic and a complex cellular microenvironment where B cells undergo repeated rounds of mutation and selection. Three major cellular components appear necessary for the GC reaction: the follicular dendritic cells (FDCs) that define the locus of GC formation and serve as antigen retaining cells for GC B cells, antigen-specific T cells, and antigen-specific B cells. GCs are divided into a dark zone (DZ) that is proximal to the T-cell area and contains rapidly dividing Ig–B cells called centroblasts, and a light zone (LZ), where non-dividing Ig+ centrocytes undergo intense selection. GC B cells take up antigen from FDC, process it and present it to antigen-specific T cells. GC T cells that recognize the antigens presented by centrocytes deliver two types of stimuli that result in the proliferation and differentiation of B cells: contact-mediated stimuli and activating cytokines. The principal molecules involved in contact-mediated B-cell stimulation are CD40 on B cells and CD40 ligand on activated T cells. Signals delivered through CD40 and CD40 ligand to each cell compartment seem to be critically involved in the regulation of GC formation; CD40-mediated signals to B cells are well-known to strongly upregulate B-cell proliferation and differentiation to apoptosis. CD40 ligand-mediated signals are also important for GC formation, presumably by controlling the activation state of GC T cells. Additionally, the differentiation of GC B cells into either memory B cells or plasma cells is dependent on whether B cells receive signals from CD40 ligand. SHM indicates somatic hypermutation. Modified and adapted from Klein & Dalla-Favera, 2008. (B) Within the GC, B cells undergo high rate proliferation and affinity maturation, are selected by antigen, switch toward advanced isotypes, and finally differentiate into either memory B cells or plasma cells. Currently, known key players at the GC stage include BCL6 and activation-induced cytidine deaminase (AID), the former necessary for GC formation while the latter is essential for class switch recombination (CSR) and somatic mutations of the immunoglobulin genes. Toward the completion of GC reaction and terminal differentiation, the expression of these genes is downregulated, while genes necessary for plasma cell formation, such as IRF4 (also known as MUM-1), PRDM1, encoding Blimp-1, and XBP1 are expressed. CD40 receptor engagement, usually caused in vivo by the interaction of GC lymphocytes with surrounding T cells at the terminal stages of GC reaction, leads to NF-κB-mediated transcriptional activation of the IRF4 gene. Involvement of the p50 and p65 subunits, which bind the IRF4 promoter, suggests that the ‘canonical,’ as opposed to the ‘alternative’ NF-κB pathway mediates this process. Furthermore, IRF4 transcriptional factor binds to the promoter region of the BCL6 gene and directly represses its transcription. Modified and adapted from Lossos, 2007.

The Germinal centre reaction starts when naive B cells are activated by antigen in the presence of costimulatory signals from T cells. These events transform B cells into centroblasts that proliferate and undergo somatic mutation of immunoglobulin genes within the ‘dark zone’. Centroblasts then develop into non-cycling centrocytes, which are selected in the light zone based on their ability to bind their cognate antigen with the help of FDCs and T cells (Fig 1) (Thorbecke et al, 1994; Han et al, 1997; Natkunam, 2007). In the GC light zone, engagement of CD40 on B cells by CD40 ligand on the T cells activates nuclear factor (NF)-κB canonical pathways and induces transcription of IRF-4 transcriptional factor. IRF4 can bind to the BCL6 promoter region and downregulates its transcription, thus allowing terminal differentiation to post-GC lymphocytes (Fig 1B) (Lossos, 2007).

B-cell trafficking between the dark and the light zone in GC physiology is regulated by a complex mechanism that is based on the interplay of specialized chemokines and their relative receptors. Centroblasts express CXC receptor 4 (CXCR4) and migrate towards a gradient of CXC-chemokine ligand 12 (CXCL12 or SDF1) originating in the light zone and mostly produced by stromal cells (Allen et al, 2004). Centrocytes instead express CXCR5, which attracts cells towards a gradient of CXCL13 produced in the dark zone. FDCs have been identified as the main cellular source of this chemokine in lymphoid organs (Ansel et al, 2000). Cycles of up and downregulation of CXCR4 and CXCR5 receptors might drive circulation of cells between the two zones.

In GC-derived lymphomas CXCL13 has also been shown to be secreted by follicular lymphoma (FL) cells, which also express CXCR4 and 5 (Husson et al, 2002). CXCL13 expression has been described in angioimmunoblastic T-cell lymphoma (AITL) cases (Dupuis et al, 2006). CXCL13 has also been recently found to be normally expressed on a major subset of FDCs and retained during their transformation in follicular dendritic cell sarcoma (FDC-S) and Castleman’s disease (Vermi et al, 2008). Noticeably, reactivity for CXCL13 is highly specific for FDC-S and appears to be associated with local accumulation of CXCR5+ tumour-associated lymphocytes. Furthermore, gene expression profiling studies in FL have raised the possibility that survival of patients with this lymphoma may be associated with ‘immune response’ signatures preferentially expressed in non-malignant T cells and macrophages and/or dendritic cells (Dave et al, 2004).

This review will examine some special categories of GC-derived lymphomas through their integrated microenvironment. We will highlight those microenvironmental findings that may serve as useful surrogate markers for tumour diagnosis or tumour progression, together with key molecules involved in tumour development.

Germinal centre-derived B-cell lymphomas

Follicular lymphoma

Follicular lymphomas are derived from GC B cells and maintain the gene expression programme of this stage of differentiation (Dave et al, 2004). Unlike normal GC B cells, roughly 85% of FLs express BCL2 as a result of the characteristic t(14;18) translocation. In some cases FL ‘transforms’ into an aggressive lymphoma resembling diffuse large B-cell lymphoma (DLBCL), and this transformation can be associated with a variety of oncogenic changes (Lossos & Levy, 2003). Accumulation of genomic alterations and clonal selection account for subsequent FL progression and transformation. However, a role for the immunological microenvironment of FL in determining clinical behaviour and prognosis of the disease has also recently been substantiated.

When FL cells are cultured in vitro, they rapidly undergo apoptosis. This finding shows that FL cells are not per se capable of maintaining an infinite lifespan and that in the in vitro condition they lack some factor, which allows instead their survival in vivo. A growing body of literature indicates that participation of bystander, non-tumoral cells plays a fundamental role both in the onset and in the progression of the disease. However, the question regarding how much of the defective apoptosis should be attributed to intrinsic defects of the neoplastic cells or to extrinsic factors that influence their behaviour has not been answered yet.

In FL the tumour cells reside and proliferate in follicular structures in close association with helper T (Th) cells and FDCs (Manconi et al, 1988), irrespective of the biopsy site (Carbone et al, 1985) (Figs 2A and 3). Therefore, the lymphoma cells seem to require the cellular interactions in the GC-like environment for their proliferation, and to retain key features of normal GC B cells including the interaction with T cells and FDCs in the follicular microenvironment (Carbone et al, 1995a).

Figure 2.

 The Germinal centre-derived lymphomas seen through their cellular microenvironment. (A) Non-neoplastic microenvironmental cells found in neoplastic follicles of follicular lymphoma. Neoplastic follicles in follicular lymphoma contain, in addition to follicular dendritic cells (FDC) (CD21+, CD23+) other non-neoplastic cells normally found in germinal centre (GC) including macrophages (CD68+) and GC T cells (CD3+, CD4+, CD57+, PD1+, and CXCL13+). (B) Non-neoplastic bystander cells in the nodules of nodular lymphocyte predominant Hodgkin lymphomas (NLPHL). Large spherical meshworks of FDC are filled with non-neoplastic bystander small CD20+ B cells and non-neoplastic CD4+ GC T cells. These T cells express markers like CD57, PD1, BCL6 and MUM1/IRF4. (C) The classic Hodgkin lymphoma (cHL) microenvironment. cHL is characterized by an infiltration of different types of cells into the lymphoma tissues, including T cells, B cells, plasma cells, neutrophils, eosinophils, mast cells, and fibroblasts. The figure shows the cells involved in direct cellular interactions with Hodgkin and Reed–Sternberg (HRS) cells. CD4+ T helper (Th)-cells are attracted by Hodgkin and RS cells through secretion of the chemokines TARC, CC chemokine 5 (CCL5) and CCL22. These, as well as CCL20, also attract TReg cells. CCL5 has an additional role in the recruitment of eosinophils and mast cells. Eosinophils may stimulate HRS cells through CD30–CD30 ligand (CD30L) interaction, whereas granulocytes may stimulate HRS cells through APRIL–BCMA interaction. The cellular interactions between CD4+ TH cells probably involve key molecules of cognate B cell–T cell interaction such as CD40–CD40L. (D) The angioimmunoblastic T-cell lymphoma (AITL) microenvironment. AILT characteristically contain a prominent proliferation of high endothelial venules (HEV) and FDC. Neoplastic cell of AITL form small clusters around FDC meshworks and HEV and are admixed with B cells, eosinophils and plasma cells.

Figure 3.

 Follicular lymphoma. (A) CD23 immunostaining depicts a sparse meshwork of FDC in a neoplastic follicle. Haematoxylin counterstain, original magnification ×400. (B) Double staining with CD23 and CD68 depicts a sparse CD23 positive meshwork of FDC (blue) in a neoplastic follicle. CD68 positive macrophages (brown) are seen around and within the dendritic meshwork. Haematoxylin counterstain, original magnification ×200.

Follicular lymphoma is a malignancy of follicle centre B cells that have at least a partially follicular pattern. The disease is characterized by a highly variable clinical course with frequent relapses. Median survival is about 10 years, but the range is very wide, with some patients surviving for 15–20 years and 10–15% of patients following a rapidly fatal course and dying within 1 year after diagnosis. Therefore, in FL there is a need for pathological predictors of outcome and response to treatment (Harris et al, 2008; Jaffe et al, 2008).

Oligonucleotide microarray-based gene expression profiling has revealed that the length of survival following diagnosis of FL can be predicted by biological differences among the tumours at the time of diagnosis (Dave et al, 2004). A training set of 95 FL patients’ biopsies was used to derive good- and poor-prognosis signatures, characterized by increased expression of a subset of genes. These findings were further validated on a test set of 96 patients. In order to discriminate whether the signatures were associated with tumour or stromal cells, CD19+ malignant cells were separated from the CD19 non-malignant cells by fluorescence-activated cell sorting, and each subpopulation was profiled again for gene expression. This experiment showed that survival of patients with FL might be associated with ‘immune response’ signatures expressed by non-malignant cells, such as T cells and macrophages (Dave et al, 2004).

The fact that FL is a malignancy primarily related to defects in the induction of apoptosis is widely accepted, judging from the clinical course and the in vivo data. Evidence from in vitro studies points to a possible role of CD40 and its interactions with CD40L in the pathogenesis of FL. This interaction plays a very important role in GC physiology. The initiation of GC response depends critically on the interactions between co-stimulatory B-cell surface receptors and ligands expressed by T cells or antigen-presenting cells. The most important of them involves the tumour-necrosis factor (TNF)-receptor family member CD40, which is expressed by all B cells, and its ligand CD40L (or CD154), expressed by Th cells.

As mentioned above, and despite constitutive high levels of bcl-2 because of the t(14;18) translocation, freshly collected FL cells rapidly undergo apoptosis in in vitro culture conditions. However, activation of malignant cells by the soluble form of CD40L results in protection from apoptosis (Husson et al, 2002), probably due to an upregulation of bcl-xL. No correlation was found between the levels of bcl-2 protein in FL cells and the spontaneous onset of apoptosis nor the CD40-mediated rescue. In contrast to previous reports (Johnson et al, 1993), no proliferation of malignant cells could be detected after CD40 stimulation. A possible explanation for these contrasting findings might be provided by the observation that CD40-dependent rescue of human B cells from apoptosis seems just to depend on a minimal cross-linking of the receptor, as it can be achieved by means of antibodies that are not specific for the CD40L epitope or by monomeric CD40L (Pound et al, 1999). However, cell cycle progression and homotypic adhesion require stimulation by soluble trimeric CD40L and simultaneous presence of CD32L-expressing transfectants. In addition to the in vitro evidence of a possible role for CD40L-bearing non-tumoral cells in FL, an abundance of polyclonal activated T cells, predominantly of the Th type (Carbone et al, 1995a; Husson et al, 2002), has been observed in involved lymph nodes. Furthermore, proliferation of malignant clonal cells occurs in response to interleukin 4 and other cytokines of T cell origin (Schmitter et al, 1997). Therefore, FL cells are not only capable of responding to physiological signals but also have the availability of such signals within their neoplastic microenvironment.

Hodgkin lymphoma (nodular lymphocyte predominant Hodgkin lymphoma, classic Hodgkin lymphoma)

Nodular lymphocyte predominant Hodgkin lymphoma (NLPHL) and classic Hodgkin lymphoma (cHL) differ in terms of clinical features, morphology, immunophenotype and expression of the B-cell programme by the neoplastic cells. Both share the unusual feature of having a few neoplastic cells in an abundant non-neoplastic cellular microenvironment including inflammatory and accessory cells (Poppema et al, 2008).

Nodular lymphocyte predominant Hodgkin lymphoma

Nodular lymphocyte predominant Hodgkin lymphoma is a monoclonal B-cell neoplasm characterized by a nodular, or a nodular and diffuse proliferation of Reed-Sternberg (RS) cell variants, known as popcorn or lymphocyte predominant cells (LP cells). These cells reside in spherical meshworks of FDCs that are filled with non-neoplastic inflammatory cells (Poppema et al, 2008). In addition to aggregates of FDCs, the background infiltrate includes small B-cells, T cells, and histiocytes. Furthermore, the nodules of NLPHL are characterized by an increase in GC-derived CD57+ T cells that are closely associated with the neoplastic LP cells (Timens et al, 1986; Poppema, 1989; Kraus & Haley, 2000; Carbone et al, 2002) (Figs 2B and 4).

Figure 4.

 Nodular lymphocyte predominant Hodgkin lymphoma (NLPHL). Two-colour immunohistochemistry reveals that a ring of MUM1-positive and CD57-positive cells is present around an unstained LP cell of NLPHL. MUM1-positive cells are stained brown in their nucleus and CD57-positive cells are stained reddish in their cell membrane. Two-colour staining, haematoxylin counterstain, original magnification ×600.

Lymphocyte predominant cells are positive for CD20, CD79a, CD75, BCL6 and CD45 in nearly all cases, and epithelial membrane antigen in more than 50% of cases. In contrast to RS cells from cHL, OCT-2, BOB.1 and activation-induced diaminase (AID) are consistently coexpressed in NLPHL. LP cells lack CD15 and CD30 in nearly all instances. However, CD30 positive large cells may be seen: these are, in most instances, reactive immunoblasts unrelated to the LP cells (Poppema et al, 2008).

As revealed by global gene expression analysis (Brune et al, 2008), LP cells show a strong similarity to the tumour cells of T cell-rich B cell lymphoma and cHL, a partial loss of their B cell phenotype, and deregulation of many apoptosis regulators and putative oncogenes. Importantly, LP cells are characterized by constitutive NF-κB activity.

Most lesions that were diagnosed in the past as diffuse LPHL were probably T-cell-rich large B-cell lymphoma. At present an overlap between NLPHL and T-cell-rich large B-cell lymphoma cannot be excluded (Poppema et al, 2008). According to current criteria the detection of one nodule showing the typical features of NLPHL in an otherwise diffuse growth pattern is sufficient to exclude the diagnosis of a primary T-cell/histiocyte-rich large B-cell lymphoma (THRLBCL) (Boudova et al, 2003). Specifically, in the differential diagnosis between NLPHL and primary THRLBCL, presence of small B cells and CD4 positive, CD57 positive or MUM1/IRF4 positive, T cells favours NLPHL, whereas presence of granzyme B positive and TIA1 positive T cells favours primary THRLBCL (Poppema et al, 2008) (Table I and Fig 5).

Table I.   Differential diagnosis of germinal centre B-cell lymphomas according to their background.
BackgroundClassic HLNLPHLTHRLBCL
  1. HL, Hodgkin lymphoma; NLPHL, nodular lymphocyte predominant Hodgkin lymphoma; THRLBCL, T cell/histiocyte rich large B-cell lymphoma.

Small B cells−/++−/+
T cells+−/++
B cell/B and T cells++
Cytotoxic TIA1+−/+−/++
Granzyme-B−/+−/++
CD4+ rosetting T cells++
CD8+ rosetting T cells
CD57+ rosetting T cells−/+
CD40L+ rosetting T cells+
MUM1/IRF4+ rosetting T cells+
Histiocytes−/++
DRC meshworks+
Figure 5.

 Nodular lymphocyte predominant Hodgkin lymphoma (NLPHL) and T-cell/histiocyte-rich large B-cell lymphoma (THRLBCL). (A) Presence in the background of CD57 positive (left) and MUM1/IRF4 positive (right) CD4 positive T cells is specific of NLPHL. (B) Presence in the background of granzyme B positive T cells (left) and numerous CD68 positive histiocytes (right) is specific for THRLBCL. Original magnification ×400 (A), ×250 (B).

Most of LP cells are ringed by CD3+ T cells, and to a lesser extent by CD57+ T cells. These T cells express markers like c-MAF, PD-1 (programmed death-1), BCL6, IRF4/MUM1, which are all consistent with a subset of GC T cells (Fig 4). CD8, TIA, and CD40L positive cells are absent whereas CD4+/CD8+ cells are relatively frequent (Atayar et al, 2007; Carbone et al, 2002; reviewed in Poppema et al, 2008).

Classic Hodgkin lymphoma

Classic Hodgkin lymphoma is a monoclonal lymphoid neoplasm derived from B cells composed of mononuclear Hodgkin cells and multinucleated RS cells residing in an abundant cellular microenvironment containing non-neoplastic small lymphocytes, eosinophils, histiocytes, plasma cells and fibroblasts (Figs 2C and 6). Based on the characteristics of the reactive infiltrate, four histological subtypes have been distinguished: lymphocyte-rich cHL, nodular sclerosis cHL, mixed cellularity cHL and lymphocyte-depleted cHL (Stein et al, 2008). In these histological subtypes the immunophenotypic and genetic features of mononuclear Hodgkin cells and RS cells are identical whereas their association with Epstein-Barr virus (EBV) show differences. In cHL EBV is more commonly associated with the mixed-cellularity subtype; NLPHL cases are very rarely associated with EBV (Khalidi et al, 1997).

Figure 6.

 Classic Hodgkin lymphoma (cHL). Cell suspension from a lymph node involved by cHL. A Reed-Sternberg cell is surrounded by CD40L positive lymphocytes. CD40L positivity is manifested as dot-like or membrane staining (left). Rosetting T cells coexpress CD4 (brown) and CD40L+ (blue) (right). Left: alkaline phosphatase-antialkaline phosphatase method, haematoxylin counterstain; Right: two-colour staining. Original magnification ×600.

Reed-Sternberg cells are positive for CD30 in nearly all cases, and for CD15 in the majority (75–85%) of cases. CD45 is usually negative (Stein et al, 2008). Regarding B-cell markers, CD40 is positive in all cases (Carbone et al, 1995b), CD20 is positive in a minority of neoplastic cells in 30–40% of cases, while CD79a is less often expressed (Stein et al, 2008). The plasma cell-specific transcription factor MUM1/IRF4 is consistently positive, usually at high intensity in RS cells (Falini et al, 2000; Carbone et al, 2002). cHL is associated with overexpression and an abnormal pattern of cytokines and chemokines and/or their receptors in RS cells that are involved in the attraction of many of the microenvironmental cells into the lymphoma background (Kuppers, 2009). Several observations indicate that RS cells are dependent on survival signals received from other cells: triggering of CD40 signalling by CD40L-expressing rosetting T cells, activation of TACI and BCMA through production of their ligand APRIL by neutrophils, and perhaps activation of CD30 through CD30L-expressing mast cells and eosinophils. A considerable fraction of infiltrating CD4+ T cells comprises regulatory T (TReg) cells, which have been shown to have immunosuppressive activity on HL-infiltrating cytotoxic T cells (for a review, see Kuppers, 2009).

Recently, it was suggested that TReg cells and PD1+ T-cells also interact with RS cells (Marshall et al, 2004; Juszczynski et al, 2007; Yamamoto et al, 2008), which produce the T-reg attractant galectin-18 and the PD-1 ligand, PDL-110. On the other hand, the observation of numerous CXCR3+ lymphocytes in some HL tumours has raised the possibility of an occasional Th1-predominant immune response (Teichmann et al, 2005).

Germinal centre-derived T-cell lymphomas

Angioimmunoblastic T-cell lymphoma

The clinical behaviour of AITL is very aggressive, the response to therapy is scarce, and the long-term outcome is dismal (Jaffe, 1995; Siegert et al, 1995; Ascani et al, 1997). The lymph node histological picture is also distinctive, constituted by a polymorphic infiltrate, a marked proliferation of high endothelial venules, and a dense meshwork of FDCs (Dogan et al, 2008; Jaffe, 2006; Went et al, 2006) (Figs 2D and 7). The polymorphous infiltrate is constituted by small- to medium-sized lymphocytes, usually with clear cytoplasm (Dunleavy et al, 2007). The lymph node architecture is largely effaced, and regressed follicles may be seen. The lymphocytes show minimal cytological atypia, and this form of lymphoma may be difficult to distinguish from atypical T-zone hyperplasia. Neoplastic cells are admixed with small reactive lymphocytes, eosinophils, plasma cells, histiocytes, and abundant FDCs (Dunleavy et al, 2007). High endothelium venules are numerous and show arborization.

Figure 7.

 Angioimmunoblastic T-cell lymphoma. Low power view of CD21 immunostain highlighting marked follicular dendritic cell proliferation entrapping high endothelial venules. Immunohistochemistry haematoxylin counterstain. Original magnification ×400.

Typically, the tumour cells of AITL have a Th cell phenotype expressing CD3, CD4, and frequently CD10, similar to a subset of normal GC-Th cells (Attygalle et al, 2002; Piccaluga et al, 2007). The neoplastic CD4+ T cells represent a minority of the lymph node cell population, their detection being facilitated by the aberrant expression of CD10. The phenotypic profile is aberrant in the majority of cases, CD5 and CD7 being the most frequently defective antigens, whereas CD10 is frequently present (Went et al, 2006). CD4 is most commonly expressed rather than CD8, but notably, these two antigens are coexpressed or even not expressed in more than 50% of cases (double-positive and double-negative cases respectively) (Went et al, 2006). The lymph nodes also contain polyclonal plasma cells, as well as frequent large B immunoblasts, despite the absence of well-formed follicles with GCs and, as noted, FDCs (Went et al, 2006). AITL is a lymphoma in which expanding B-cell clones are often present beside the T-cell clones. In early lymph-node involvement by AITL, the neoplastic T cells preferentially occupy the B-cell follicles and immediate perifollicular area, sometimes mimicking a FL of B-cell origin. This suggests that the follicle-GC microenvironment is critical for tumour development (Attygalle et al, 2002; Dorfman et al, 2006; de Leval et al, 2007).

A great majority of AITL cases harbour an EBV-infected B-cell population. This pathogenetic characteristic is common to other B-cell malignancies originating from the GC (discussed below), which include Burkitt lymphoma and cHL [reviewed in Carbone et al, 2008]. In AITL, virus presence can be detected in B lymphocytes as well as in B immunoblasts. It is possible that reactivation of EBV could be consequent to a decreased immune surveillance in the setting of a compromised immune system (Weiss et al, 1992), although EBV-positive B cells can be found very early during the course of the disease. In some cases, EBV infection of B cells progresses to such a state as to resemble to a B-cell lymphoproliferative disorder similar to a post-transplant polymorphic B-cell lymphoma. In other patients, EBV-positive B cells assume morphological characteristics of RS cells, leading to an erroneous diagnosis of cHL. The virus is only rarely seen in T cells, therefore giving force to the suggestion that EBV infection does not contribute to the aetiology of the disease (Quintanilla-Martinez et al, 1999).

The molecular pathogenesis of AILT, as for all peripheral T-cell neoplasms in general, is poorly understood. Recent studies have made use of gene expression profiling to strongly support the view that normal follicular Th cells (TFH) represent the normal counterpart of AITL (de Leval et al, 2007). The molecular profile of AITL is characterized, in addition to a strong microenvironment imprint (overexpression of B-cell- and FDC-related genes, chemokines, and genes related to extracellular matrix and vascular biology), by overexpression of several genes characteristic of normal TFH cells, such as CXCL13, BCL6, PDCD1, CD40LG and NFATC1. Overexpression of these genes in AITL was validated by immunohistochemistry (de Leval et al, 2007). In another study, additional markers for GC T cells, such as PD-1 and SAP, were found to be expressed in 95% of AITL specimens examined (Roncador et al, 2007). The possible pathogenetic role of some of these markers will be described in the following paragraphs.

Common pathogenetic aspects

In this section we will discuss some common aspects of the pathogenesis of GC-derived lymphomas.

Epstein-Barr virus has been associated with dysregulation of GC B cells and is closely linked with a number of lymphoid neoplasms that include Burkitt lymphoma and cHL, post-transplant lymphoproliferative disorders and AITL (discussed above). The influence of EBV infection on the microenvironment remains unclear. Independent studies have recently demonstrated that EBV can transform antigen receptor–deficient GC B cells and enable their escape from apoptosis. The continued survival of the ‘rescued’ pre-apoptotic B cells allows their proliferation. The EBV-encoded latent membrane protein (LMP) 2A is likely to function as the surrogate receptor through which B-cell signalling is triggered (Bechtel et al, 2005; Chaganti et al, 2005; Mancao et al, 2005). This mechanism of EBV/LMP2A-induced escape of antigen receptor–deficient GC B cells from apoptosis offers an intriguing model of lymphomagenesis.

SAP, or SLAM-associated protein, plays a central role in the regulation of Signalling Lymphocyte Activation Molecule (SLAM). SLAM is a protein that is centrally involved in the bidirectional stimulation of T and B cells. When activated, it mediates expansion of activated T cells during immune responses, induces production of interferon-gamma, and changes the functional profile of subsets of T cells. Signalling through SLAM-SLAM binding during mutual interaction between B cells, and between B cells and T cells, increases the expansion and differentiation of activated B cells. SAP is thought to be involved in the coordination of the immune response to EBV or other viral infections and is expressed in natural killer (NK), CD4, and CD8 T cells but not in monocytes and primary B cells, although expression in certain B-cell lines has been documented (Parolini et al, 2003; Nichols et al, 2005). The high levels of SAP in AITL are in keeping with the low frequency of EBV infection in T cells.

Among the transcription factors associated with physiological GC events whose expression is altered in FLs, a potential role has emerged for interferon regulatory factor 8 (IRF8). IRF8 is a member of the IRF family of transcription factors whose members play critical roles in interferon (IFN) signalling pathways governing the establishment of innate immune responses by myeloid and dendritic cells (Wang & Morse, 2008). IRF8 is also expressed in lymphoid cells and recent studies have documented its involvement in B cell lineage specification, immunoglobulin light chain gene rearrangement, the distribution of mature B cells into the marginal zone and follicular B cell compartment, and the transcriptional regulation of critical elements of the GC reaction (Wang & Morse, 2008; Wang et al, 2008). Martinez et al (2008) studied the expression of IRF8 in reactive lymphoid tissues and in a series of 232 B cell tumours and 30 cell lines representing a variety of B cell developmental stages. Although IRF8 was detectable in most reactive B cells, its expression levels differed with developmental stage. GC B cells contained the highest levels of IRF8, with lower levels seen in mantle and marginal zone B cells and none in plasma cells. IRF8 was coexpressed with PAX-5, Pu.1, and BCL6 and, similar to BCL6, was absent from the small population of IRF4-positive GC B cells thought to be committed to postgerminal centre developmental programmes. Similarly, IRF8 was most strongly expressed in lymphomas of GC origin (FLs), with somewhat lower levels present in mantle cell lymphomas, chronic lymphocytic leukaemia and marginal zone lymphomas, and no expression observed in plasmacytic/plasmablastic neoplasms. The reciprocal expression pattern with IRF4 in reactive tissues was generally maintained in lymphomas with some exceptions. These results provide a new diagnostic marker that is helpful in distinguishing B cell non-Hodgkin lymphoma (NHL) subtypes (Martinez et al, 2008).

From a pathogenetic point of view, although a clear role for CXCL13 production in distinct lymphomas has not yet been identified, it might point to the need for specific chemokine/receptor interaction, leading not only to neoplastic cell accumulation into secondary lymphoid tissues, but also to the typical architectural organisation of the disease. The hypothesis has been formulated that CXCL13 production by lymphoma cells may act as a chemoattractant for CXCR5+ cells, thereby contributing to the capacity of malignant cells to create a microenvironment necessary for their survival (Husson et al, 2002).

PD-1, a member of the CD28 costimulatory receptor family, is expressed by GC-associated T cells in reactive lymphoid tissue (Riley & June, 2005). In a study of a wide range of lymphoproliferative disorders, neoplastic T cells in 23 cases of AITL were immunoreactive for PD-1, while other subtypes of T cell and B cell NHL, as well as cHL, did not express PD-1 (Dorfman et al, 2006). The pattern of PD-1 immunostaining of neoplastic cells in AITL was similar to that reported for CD10, a marker of neoplastic T cells in AITL. Tumour-associated FDCs in cases of AITL were found to express PD-L1, the PD-1 ligand (Dorfman et al, 2006). These findings reinforce the hypothesis that AILT is a neoplasm of GC-associated T cells. PD-1 is also a marker of follicular T cells that form rosetting around the neoplastic cells in NLPHL. Nam-Cha et al (2008) analyzed the diagnostic value of PD-1 staining in 152 cases of various types of HL and THRLBCL. Results show that PD-1 is a marker not only of follicular T cell rosettes in NLPHL, but also of a subset of lymphocyte-rich cHL and is absent in THRLBCL. The presence of PD-1-positive T cell rosetting, therefore, seems to be an additional useful feature in the, sometimes difficult, differential diagnosis of NLPHL and T/HRBCL. PD-1 expression was also reported in some T cell NHL subtypes. Xerri et al (2008) described the expression profile of PD-1 and its ligands (PD-L1 and PD-L2) in 161 lymphoma tissue and 11 blood samples of B-NHLs. In reactive lymph nodes, PD-1 was mainly expressed in follicular T cells. In B-NHLs, PD-1 was mainly expressed in reactive T cells as well, but expression was also noted in neoplastic B cells from small lymphocytic lymphoma (SLL, 12/13), grade III FL (3/3), and DLBCL (2/25). In contrast, neoplastic B cells from MCL (0/11), marginal zone lymphoma (0/12), BL (0/3), and grade 1 to 2 FL (0/40) were PD-1 negative. PD-L1 and PD-L2 were negative in small B cell lymphomas, including B-SLL. Flow cytometry showed that blood cells from chronic lymphocytic leukaemia (BCLL) also expressed PD-1, which could be increased by CD40 stimulation (Xerri et al, 2008). The above results suggest that frequent PD-L1 (and PD-L2) positivity in malignant cells from DLBCL and HL points to a possible functional mechanism of immune escape using PD-1 triggering of intra-tumoural T cells. This hypothesis is consistent with previous data suggesting that the expression of PD-L1 may serve as a mechanism for potentially immunogenic tumours to escape from host immune responses (Blank et al, 2005). There might also be potential therapeutic implications, as blocking of the interaction between PD-1 and PD-L1/PD-L2 has been suggested as a promising strategy for specific tumour immunotherapy (Hirano et al, 2005; Iwai et al, 2005).

Microenvironmental cells and tumour progression

This section discusses the interaction of GC-derived lymphomas with the tumour microenvironment, together with the possible role of certain microenvironmental cells in tumour progression.

Immunohistological analysis in FL showed that increased infiltrating lymphoma-associated macrophages conferred a worse overall survival (Farinha et al, 2005). Similarly, tumour-infiltrating lymphocytes expressing CD4 and FoxP3 have been shown to correlate with improved survival in patients with FL (Alvaro et al, 2006; Carreras et al, 2006; Lee et al, 2006). Furthermore, another immunophenotypic study compared the expression of FDC markers CD21, CD23, CD35, CXCL13, the stromal markers low-affinity nerve growth factor receptor (LNGFR) and CNA.42 in 35 FL cases with reactive lymphoid tissue. CXCL13 was expressed by follicular stroma in all FLs, but most cases showed either partial or complete absence of other FDC antigens. Only a minority of FL cases (14/34, 40%) showed stroma that resembled mature FDCs (CD23+, CD21+, CD35+) and these tumours were always associated with numerous intrafollicular T cells, similar to reactive GCs. Serial biopsy specimens analyzed in a subset of 15 patients showed loss of FDC antigens in tumour-associated stroma over time. According to this study, the pattern of FDC loss in stroma of FL may serve as a useful surrogate marker of tumour progression (Chang et al, 2003).

Interestingly, a recent oligonucleotide microarray-based study profiled gene expression in DLBCL, identifying signatures predictive of good and bad survival rates respectively (Lenz et al, 2008). The prognostically favourable signature called ‘stromal 1’ includes genes encoding for extracellular matrix proteins and identifies tumours with vigorous extracellular-matrix deposition and infiltration by cells of the monocytic lineage. The prognostically unfavourable ‘stromal 2’ signature includes angiogenetic switch-related genes (the therapeutical application of this study is discussed below).

Several retrospective studies using immunohistochemistry (IHC) have attempted to define adverse prognostic markers associated with RS cells, such as high expression of BCL2 (Sup et al, 2005) or topoisomerase-IIα (topo-IIα) (Doussis-Anagnostopoulou et al, 2008), and/or loss of human germinal centre-associated lymphoma (Natkunam et al, 2007) or human leucocyte antigen (HLA) Class II molecules (Diepstra et al, 2007). Other IHC reports have highlighted the characteristics of non-malignant immune cells that may predict unfavourable outcome, in particular a low infiltration of intratumoural FOXP3+ TReg cells in combination with a high percentage of either granzyme-B+ or TIA-1+ lymphocytes (Alvaro et al, 2005; Chetaille et al, 2008).

Conclusions

In the scenario of GC-derived lymphomas, tumour progression probably depends on signals mediated by cellular interactions in the lymphoma microenvironment more than on transforming events. The dependency of lymphomas on microenvironmental survival signals may contribute to the close similarity between GC-derived lymphomas and the normal GC counterparts in terms of phenotype and gene expression pattern. New treatment options should be developed on the basis of the current knowledge of the role of the microenvironment in lymphoma progression by interfering with survival or proliferative signals from other cells in the lymphoma environment.

An interesting therapeutic approach in cancer patients is the combined blockade of epidermal growth factor receptor (EGFR) and vascular endothelial growth factor (VEGF) receptor signalling, which both contribute to tumour progression. In this regard, different tyrosine kinase inhibitors that are able to block the activation of both receptors have been synthesized and are in clinical development (De Luca et al, 2008). EGFR antagonists are currently available for the treatment of specific metastatic epithelial cancers. Less information is available regarding the use of EGFR antagonists in the treatment of lymphomas. On the other hand, the monoclonal antibody to VEGF, bevacizumab, is currently being investigated in several clinical trials involving patients with DLBCL. It is possible that those DLBCL patients with an increased tumour blood-vessel density/stromal-2 signature (Lenz et al, 2008) may benefit from this angiogenesis inhibitor. Moreover, the heavy infiltration of some lymphomas with myeloid-lineage cells (Lenz et al, 2008) raises the possibility that monoclonal antibodies targeting antigens on the myeloid-lineage cells could interfere with microenvironmental interactions between these cells and lymphoma cells.

Acknowledgements

This work was supported in part by a grant from the Ministero della Salute, Rome, within the framework of the Progetto Integrato Oncologia-Advanced Molecular Diagnostics ‘Multidimensional characterization of tumours’ project (to AC). GE is a Conway Fellow; support from the UCD Conway Mass Spectrometry Resource is acknowledged.

Ancillary