Epidemiological, biological and clinical features of HIV-related lymphomas in the era of highly active antiretroviral therapy.

Abstract

Introduction The association between HIV infection and the development of lymphoma was noted in the early phases of the AIDS epidemic. Since the outbreak of AIDS in 1981, the incidence of AIDS-related non-Hodgkin's lymphomas (AIDS–NHL) has been increasing steadily, and in 1986 the Centers for Disease Control recognized NHL as an AIDS-defining illness [1]. In the era preceding the introduction of highly active antiretroviral therapy (HAART), NHL represented the second most frequent cancer associated with AIDS after Kaposi's sarcoma (KS). In contrast to KS, which predominated in homosexual individuals, AIDS–NHL distributed homogeneously throughout the spectrum of HIV risk groups, including haemophiliac patients in whom KS was exceeded by AIDS–NHL. In addition, whereas KS was generally an early manifestation of HIV infection, most AIDS–NHL tended to develop relatively late in the natural history of the disease. More recently, the epidemiological differences between AIDS–NHL and KS have been reinforced by the striking reduction in the incidence of KS in HAART-treated patients, whereas the frequency of systemic AIDS–NHL has remained substantially the same despite more effective antiretroviral therapy [2]. Early studies of AIDS–NHL showed a number of distinctive features that are shared by these lymphomas. First, nearly all AIDS–NHL derive from B cells, are characterized by extreme clinical aggressiveness, and display a predilection for unusual locations otherwise rarely implicated in lymphomas. Despite these similarities, however, it was soon realized that AIDS–NHL are markedly heterogeneous both histologically and clinically [3–5]. Genetic studies have subsequently revealed that such clinicopathological heterogeneity reflects variability in the molecular lesions associated with these lymphomas [3,4]. In recent times, the field of AIDS-related lymphomas has been influenced by several novel acquisitions. First, biological studies have led to the identification of novel clinicopathological types of AIDS-related lymphomas, which were previously unrecognized as independent diseases [6,7]. Second, epidemiological studies have clarified that HIV-infected individuals develop an excess of Hodgkin's disease (HD) in addition to NHL, and that AIDS-related HD (AIDS–HD) is biologically and clinically distinct from HD of the immunocompetent host [8–10]. Finally, the introduction of HAART has substantially modified the approach to AIDS-related lymphomas. In fact, the reduced morbidity of AIDS patients brought about by HAART justifies the use of aggressive antineoplastic therapies for those patients who would otherwise experience a prolonged life expectancy once the lymphoma is controlled. The clinical employment of such aggressive therapies is made possible by the relative immunocompetence of HIV patients treated with HAART. On these grounds, we propose a clinical and biological framework for the management of AIDS-related lymphomas, and discuss the challenges faced by physicians who are currently called to care for HIV-infected patients with lymphoma. Systemic non-Hodgkin's lymphomas Patients with HIV infection are at increased risk of developing systemic NHL. These lymphomas are most commonly high or intermediate grade, showing aggressive clinical behaviour and poor outcome. Epidemiology In developed countries, systemic NHL account for approximately 3–4% of primary AIDS-defining illnesses [11,12]. However, approximately 5% of HIV-seropositive individuals will develop a systemic NHL as a secondary diagnosis after another AIDS-defining condition [13]. Whereas the incidence of KS has been declining after the introduction of HAART, the number of patients developing systemic AIDS–NHL has remained relatively stable [2]. Because many systemic AIDS–NHL are a late complication of HIV infection, it may be predicted that longer life expectancy of HIV-infected individuals may increase the cumulative risk of these lymphomas. All subtypes of systemic AIDS–NHL are increased in the context of AIDS, although the relative risk (RR) for low-grade lymphomas is only 14 whereas the RR for high-grade lymphomas is over 300 [14]. Pathology and molecular biology Systemic AIDS–NHL are a heterogeneous group of malignancies displaying a B cell phenotype. The overwhelming majority of systemic AIDS–NHL fall within two main histological categories [15] : small non-cleaved cell lymphoma (SNCCL), which includes classic Burkitt's lymphoma (BL) (Fig. 1) and Burkitt-like lymphoma (40%); and diffuse large cell lymphoma (DLCL), which includes large non-cleaved cell lymphoma (LNCCL) (25%), immunoblastic lymphoma plasmocytoid (IBLP) (Fig. 2) (25%) and CD30-positive anaplastic large B cell lymphoma [3,16,17]. A frequently encountered pathological characteristic of systemic AIDS–NHL is the occurrence of cases featuring a certain degree of overlap between established histological categories [18]. The best example is represented by the so-called intermediate lymphoma, which exhibits features ‘intermediate’ between SNCCL and IBLP [19–21]. Deceptively, the atypical morphological features of intermediate lymphomas hamper a correct discrimination between SNCCL and IBLP, which is clinically relevant because of the different behaviour and prognosis of these lymphomas.Fig. 1.: Burkitt's lymphoma (small non-cleaved cell lymphoma). Medium-sized tumour cells have a monotonous appearance; they show round nuclei containing multiple centrally located nucleoli. Haematoxylin-eosin stain, × 320.Fig. 2.: Immunoblastic lymphoma plasmacytoid. Immunoblasts with plasmacytic features have abundant cytoplasm and large, solitary nucleoli. Haematoxylin-eosin stain, × 400.The molecular pathogenesis of systemic AIDS–NHL is complex and has been studied in depth in the case of AIDS–BL and AIDS–DLCL [4]. In the case of AIDS–BL, the role of antigen stimulation and selection is documented by immunogenotypic studies as well as by the antigenic specificity of the B cell receptor molecules expressed by this lymphoma [4]. Viral infection of AIDS–BL tumour cells is mainly represented by Epstein–Barr virus (EBV) infection, which is restricted to approximately 30% of cases [22]. EBV infection in AIDS–BL, as well as in other AIDS–NHL, is generally monoclonal, consistent with the hypothesis that the virus has been present in the tumour progenitor cell since the early phases of its clonal expansion, and thus putatively contributed to the development of lymphoma [22]. The precise role of EBV in AIDS–BL pathogenesis, however, has remained controversial. On the one hand, EBV infection is considered to be a predisposing factor for the subsequent development of lymphoma in the context of HIV-related persistent generalized lymphadenopathy [23]. On the other hand, EBV-positive AIDS–BL express the viral latency I programme and, therefore, fail to express the EBV transforming antigens EBNA-2 and LMP-1, which are key inducers of the transformed phenotype in other B cell models [24,25]. Recently, it has been hypothesized that cell transformation in tumours expressing the EBV latency I phenotype may be mediated, at least partly, by a class of non-coding, although highly expressed small messenger RNA termed Epstein–Barr virus-encoded RNA (EBER) [26]. Because EBER are expressed in AIDS–BL, it is possible that these mRNA play a pathogenetic role in the development and growth of this lymphoma [24–26]. The profile of the molecular lesions of AIDS–BL includes activation of the c-MYC proto-oncogene in 100% of cases, inactivation of p53 in 60% of cases and point mutations of BCL-6 in 60% of cases [22,27] (Table 1).Table 1: Molecular lesions of AIDS-related non-Hodgkin's lymphomas The molecular pathogenesis of AIDS–DLCL is more heterogeneous than that of AIDS–BL [4]. In particular, genetic studies performed to date have failed to reveal a common genetic alteration in these lymphomas (Table 1). Infection by EBV occurs in a substantial proportion of cases of AIDS–DLCL [22,24,25]. However, only some of the infected cases, mainly those categorized as IBLP, express the EBV-encoded LMP-1 protein [24,25]. Conceivably, in EBV-infected AIDS–DLCL expressing LMP-1, the virus plays a major pathogenetic role. Apart from EBV infection, the only genetic alteration associated with AIDS–DLCL is represented by rearrangements and mutations of the BCL-6 proto-oncogene. Because BCL-6 mutations represent a genotypic marker of B cell transit through the germinal centre, it is assumed that AIDS–DLCL are histogenetically related to germinal centre B cells in the majority of cases [25,27]. Recent data have suggested that systemic AIDS–NHL may be segregated into two distinct phenotypical categories on the basis of the expression pattern of LMP-1, the zinc finger transcription factor BCL-6, and the proteoglycan CD138/syndecan-1. In normal B cells, BCL-6 is selectively expressed by germinal centre B cells, whereas CD138/syndecan-1 clusters with the late stages of B cell differentiation [25]. AIDS–NHL expressing the LMP-1−/CD138−/BCL-6+ phenotype include AIDS–BL (Fig. 3) and AIDS–LNCCL, whereas AIDS–NHL expressing the LMP-1+/CD138+/BCL-6− phenotype are predominantly represented by AIDS–IBLP [25] (Fig. 4).Fig. 3.: Burkitt's lymphoma (small non-cleaved cell lymphoma), displaying the BCL-6+ phenotype. Most neoplastic cells show strong nuclear immunoreactivity with the anti-BCL-6 monoclonal antibody. Paraffin-embedded tissue section, alkaline phosphatase antialkaline phosphatase method, haematoxylin counterstain. × 250.Fig. 4.: Immunoblastic lymphoma plasmacytoid, displaying the syn-1+ phenotype. Most immunoblastic-plasmacytoid tumour cells show strong cytoplasmic immunoreactivity with the anti-syn-1 monoclonal antibody. Paraffin-embedded tissue section, alkaline phosphatase antialkaline phosphatase method, haematoxylin counterstain. × 400.Only a few cases of T cell AIDS–NHL have been reported so far. They show HIV p24 expression within the tumour-associated transformed T cells or macrophages [28,29]. However, the direct role of HIV in the pathogenesis of these lymphomas remains unclear. Clinical features and therapy Typically, patients with systemic AIDS–NHL present with widespread disease and extranodal involvement at diagnosis, the most common sites being the central nervous system (CNS), gastrointestinal tract, bone marrow and liver [5,30–33] (Table 2). Several prognostic factors predictive of survival have been identified, including immune function (i.e. CD4 cell count), a previous AIDS diagnosis, Karnofsky performance status, age, lactate dehydrogenase level and response to therapy [5] (Table 3).Table 2: Clinicopathological features of 96 HIV-related systemic non-Hodgkin's lymphoma Table 3: Classical prognostic factors on survival time at univariate analyses of 96 HIV-related non-Hodgkin's lymphoma The treatment of systemic AIDS–NHL is a great challenge. Traditionally, low-dose chemotherapy has been considered the standard treatment for all patients [34]. However, although the response rate is high, the median survival of these patients is approximately 6–10 months, and only 10–20% survive free of disease for longer than 2 years. Since the advent of HAART, the prolongation of life expectancy and reduction of opportunistic infections of HIV-infected patients have called into question the value of a conservative approach with low-dose chemotherapy. Within the European Intergroup Study NHL–HIV, a randomized study has stratified patients according to the presence or absence of adverse prognostic factors (previous AIDS, CD4 cell count < 100/mm3, performance status > 1). Patients with no adverse prognostic factors (low-risk group) were randomly assigned between the intensive regimen of doxorubicin, cyclophosphamide, vindesine, bleomycin, prednisone (ACVBP) and cyclophosphamide, doxorubicin, vincristine, prednisone (CHOP) with granulocyte–colony stimulating factor (G–CSF) support, whereas patients with only one adverse prognostic factor (intermediate-risk group) were randomly assigned between CHOP and CHOP reduced to 50%[35]. High-risk patients (more than two adverse prognostic factors) were treated with CHOP reduced to 50% or vincristine plus prednisone [35]. Since the availability of HAART, patients were treated concomitantly with chemotherapy and HAART. The preliminary results of patients in the low-risk group showed a similar complete remission (CR) rate (65% in the ACVBP arm versus 56% in the CHOP arm) and a similar survival rate (51 and 43%, respectively, at 2 years) in both arms, although toxicity was higher with ACVBP. In the intermediate-risk group, the CR rate in the CHOP arm (59%) was significantly superior to that of reduced CHOP (35%;P = 0.005), but it has not yet translated into a better overall survival rate (35 versus 28% at 2 years, P = 0.09). However, the disease-free survival rate was significantly better in the CHOP arm (33 versus 25%, P = 0.04). These results seem to indicate that CHOP, the standard regimen for high-grade NHL of the general population, is also the most effective regimen for systemic AIDS–NHL, and that it should be considered standard therapy in selected HIV-positive patients. As HAART would otherwise allow a long life expectancy, the high relapse rate of AIDS–NHL represents a novel challenge. Several patients with systemic AIDS–NHL are therefore now being considered for second-line chemotherapy. Within the Italian Cooperative Group on AIDS and Tumors (GICAT), we initially treated 21 patients with resistant or relapsed AIDS–NHL with the etoposide, mitoxantrone and prednisustine regimen, which has previously been employed in elderly (≥ 70 years) patients with NHL, with acceptable toxicity and good activity. Objective responses were obtained in seven out of 19 (37%) evaluable patients. In particular, objective responses were obtained in two out of 12 evaluable patients with resistant NHL with one CR, whereas objective responses were obtained in five out of seven evaluable patients with relapsed NHL, with four CR. The median survival of patients with relapsed AIDS–NHL was 7 months, significantly longer (P = 0.02) than that of patients with resistant AIDS–NHL (2 months) [36]. In order to ameliorate these results in resistant patients, we subsequently decided to treat patients with resistant AIDS–NHL with the cyclophosphamide, doxorubicin and etoposide (CDE) infusional regimen that has shown high activity as a first-line therapy in AIDS–NHL [37]. As of December 1998, we have employed CDE in 27 patients with resistant AIDS–NHL with satisfactory preliminary data: objective responses were obtained in seven out of 23 (31%) evaluable patients, with two CR. On this basis, we suggest that relapsed AIDS–NHL may be treated with chemotherapy aimed at control of the disease, whereas resistant AIDS–NHL may be treated only with palliative intent or included in experimental approaches. Primary central nervous system lymphoma Primary central nervous system lymphoma (PCNSL) represents a late complication of HIV infection. With the introduction of HAART, the incidence of this lymphoma has dramatically increased. It derives from B cells and is represented by DLCL with a consistent infection of the tumour clone by EBV. Epidemiology In contrast to systemic AIDS–NHL, the epidemiology of AIDS–PCNSL has changed substantially since the introduction of HAART. In the pre-HAART era, PCNSL developed in 2–13% of AIDS patients during their disease with a RR of approximately 1000. Because AIDS–PCNSL generally occurred very late in the natural history of AIDS and was typically associated with very low CD4 cell counts, this lymphoma was the AIDS-defining illness in only 0.6% of HIV-infected individuals [38–41]. After the introduction of HAART, the incidence of PCNSL has decreased drastically [2] (Fig. 5). The reasons for such a decrease are not totally clear, although preserved immune function may account at least partly for this phenomenon.Fig. 5.: Non-Hodgkin's lymphoma incidence by half year of observation among 17 978 HIV-infected persons observed from January 1994 to June 1997. The decrease of incidence was statistically significant only for PCNSL (P = 0.003). - - - - - - BL, Burkitt's lymphoma; – – – – IBL, immunoblastic lymphoma; — PCNSL, primary central nervous system lymphoma. From Jacobson et al. [2].Pathology and molecular biology All AIDS–PCNSL derive from B cells and are represented by DLCL, either of the IBLP or the LNCCL type [42–44]. At the molecular level, AIDS–PCNSL are characterized by the consistent infection of the tumour clone by EBV [44,45] (Table 1). Approximately 50% of cases express the EBV-encoded transforming protein LMP-1, suggesting a direct role of the virus in the pathogenesis of these lymphomas [44]. AIDS–PCNSL also frequently express high levels of the BCL-2 protein, consistent with the well-known ability of LMP-1 to induce sustained levels of BCL-2 [43,44]. Analogous to systemic AIDS–NHL, AIDS–PCNSL may also be segregated into two phenotypic categories on the basis of the expression pattern of LMP-1, BCL-2 and BCL-6 [44]. AIDS–PCNSL expressing the LMP-1+/BCL-2+/BCL-6− phenotype generally display an IBLP morphology, whereas cases displaying the LMP-1−/BCL-2−/BCL-6+ phenotype display a LNCCL morphology [44]. Preliminary reports suggest that the phenotype of AIDS–PCNSL might be of prognostic relevance [46]. Apart from EBV infection, the only molecular lesion detectable in AIDS–PCNSL is represented by mutations of BCL-6 5’ non-coding sequences, which occur in approximately 60% of cases [44] (Table 1). Because BCL-6 mutations are acquired during B cell transit through the germinal centre [47], their occurrence in AIDS–PCNSL suggests that these tumours are histogenetically related to germinal centre B cells, which have subsequently migrated to the CNS [44]. Despite earlier suggestions, it is now believed that human herpesvirus 8 (HHV-8) is not pathogenetically related to AIDS–PCNSL [48,49]. Clinical features and therapy A major clinical problem with AIDS–PCNSL is their correct identification, because several CNS infections may antedate the diagnosis of AIDS–PCNSL. When a computed tomography scan or magnetic resonance imaging identify an intracranial mass in an HIV-infected individual, the patient is usually first treated with antitoxoplasmosis therapy, and a brain biopsy is considered only after antibiotic failure. Unfortunately, neurological deterioration occurs in non-responders, who frequently become ineligible for brain biopsy or adequate therapy. On this basis, it has been agreed that an early brain biopsy should be routinely considered for any patient with negative toxoplasmosis serology or for those who worsen within the first week of antitoxoplasmosis treatment. In order to avoid the need for a brain biopsy, a recent experience with positron emission tomography and tallium single photon emission computed tomography imaging suggested that these non-invasive techniques may be able to distinguish tumour from CNS infection and offer the opportunity of rapid diagnosis [50,51]. The predictive value of positron emission tomography and single photon emission computed tomography is further enhanced by the molecular analysis of EBV DNA sequences in the cerebrospinal fluid, because positive detection is a highly sensitive and specific marker of AIDS–PCNSL [50,51]. The prognosis of AIDS–PCNSL is influenced by performance status and the extent of disease at the time of diagnosis, including the degree of neurological dysfunction. Standard therapy is still radiation therapy alone, and it currently represents the most common therapeutic modality despite a median survival of only 2–3 months. Long-term responses are occasionally seen in those patients in whom therapy is instituted early in the course of AIDS–PCNSL [52–54]. Because of this poor outcome, there is an urgent need to try new experimental approaches and make an effort to treat patients at an earlier stage. Experimental trials combining chemotherapy and radiotherapy are ongoing, and recent data have shown a synergistic activity of methotrexate and zidovudine in the treatment of AIDS–PCNSL [55]. Finally, the role of HAART should be better evaluated in this setting. Hodgkin's disease HD represents the most common type of non-AIDS-defining tumour that occurs in the HIV population. All published series have documented unusually aggressive tumour behaviour, including a higher frequency of unfavourable histological subtypes, advanced stages and poorer therapeutic outcome compared with the behaviour of HD outside the HIV setting. Epidemiology AIDS–HD is the most common non-AIDS-defining tumour that occurs in the HIV population. RR for HD in HIV-infected individuals is very consistent in different studies, including recent linkage studies of AIDS and Cancer Registries. It was 8.8 [95% confidence interval (CI) 5.0–14.3] in a linkage study from San Francisco (USA) [56]; 8.5 (95% CI: 4.1–16) in a study from Australia [57]; 7.6 (95% CI: 4.1–13.1) in large linkage study including seven regions of the United States and Puerto Rico [58]; and 8.9 (95% CI: 4.4–16.0) in the only European study, from Italy [59]. The highest RR was found in the years around AIDS diagnosis [59], suggesting that the probability of HD is proportional to the degree of immunosuppression. These findings are in agreement with several reports demonstrating an excess of HD in HIV-seropositive cohorts [8,9], in never-married men aged 25–54 years [60], and in HIV-seropositive black patients in South Africa [61]. An association between HD and HIV infection thus seems to be well established, although with a RR much lower than that for NHL [13]. The excess of HD observed in the context of AIDS is in contrast to the lack of an increase of the disease in transplant recipients [62]. The reason for this discrepancy is not known. Pathology and molecular biology AIDS–HD exhibits pathological features that are different from those found in HD in the general population [63–65]. AIDS–HD is characterized by the predominance of unfavourable histological subtypes, namely mixed cellularity and lymphocyte depletion, frequently with an abundance of Reed–Sternberg cells (Fig. 6), which is otherwise unusual in HIV-negative patients [10,66]. The frequency of nodular sclerosis AIDS–HD is substantially lower than that observed in HIV-negative individuals and lymphocyte predominance AIDS–HD is exceptional. The cellular background of AIDS–HD is characterized by fibrohistiocytoid stromal cell proliferation and by the admixture of scarce amounts of reactive components with high numbers of neoplastic cells. This latter feature may result in histopathological patterns encompassing obvious cases of HD and cases of NHL displaying anaplastic large cell populations.Fig. 6.: Hodgkin's disease. CD30+ Reed–Sternberg cells are numerous in this field. Paraffin-embedded tissue section, avidin biotin peroxidase complex method, haematoxylin counterstain. × 250.AIDS–HD is characterized by a high frequency of EBV association (80–100%) (Fig. 7) when compared with HD in the general population [10]. The EBV genomes in AIDS–HD have been reported to be episomal and clonal [67]. The pathogenetic role of EBV in AIDS–HD is supported by the Reed–Sternberg cell expression of EBV-transforming proteins, namely LMP-1 (Fig. 8). Recent evidence suggests that LMP-1 is a key pathogenetic feature of AIDS–HD, and that its expression by Reed–Sternberg cells may account for several differences between AIDS–HD and HD in the general population [67].Fig. 7.: Hodgkin's disease. Reed–Sternberg cells show Epstein–Barr virus positivity by Epstein–Barr virus-encoded RNA in-situ hybridization. The signal is present as dense brownish grains over the nuclei of Reed–Sternberg cells. Paraffin-embedded tissue section, in-situ hybridization, haematoxylin counterstain, × 250.Fig. 8.: Hodgkin's disease. Most Reed–Sternberg cells display strong cytoplasmic staining for the Epstein–Barr virus-encoded latent membrane protein-1. Paraffin-embedded tissue section, alkaline phosphatase antialkaline phosphatase method, haematoxylin counterstain. × 400.Clinical features and therapy One of the most peculiar features of AIDS–HD is the widespread extent of the disease at presentation and the frequency of systemic ‘B’ symptoms (fever, night sweats, weight loss). At diagnosis, more than 70% of patients have ‘B’ symptoms and 85–90% have advanced-stage (III–IV according to Ann Arbor staging classification) disease, with frequent involvement of extranodal sites, the most common being bone marrow, liver and spleen. Similarly to AIDS–BL, and in contrast to AIDS–DLCL, AIDS–HD tends to develop as a relatively early manifestation of HIV infection, with a relatively high median CD4 cell count, ranging from 275 to 306/mm3[10]. The optimal therapy for AIDS–HD is controversial. Because most patients have had advanced disease, the most commonly employed protocols have included chemotherapy regimens used in the general population such as mechlorethamine, vincristine, procarbazine and prednisone (MOPP) or doxorubicin, bleomycin, vinblastine and dacarbazine (ABVD), or MOPP/ABVD. However, the CR rate in AIDS–HD is far below the usual percentage observed in HIV-negative patients, the tolerance of chemotherapy is poor, and a reduction of doses or a delay of chemotherapy are often needed, resulting in a median overall survival of approximately 1.5 years [10,66,68,69] Because approximately 85–90% of patients with AIDS–HD have advanced-stage disease, there are no data on the impact of radiotherapy in the treatment of HD in the HIV setting. It is currently thought that the outcome of AIDS–HD might improve with optimal combinations of antineoplastic and antiretroviral treatments, and that the availability of HAART might improve the control of underlying HIV infection also during chemotherapy. The inclusion of haematopoietic growth factors in the treatment of these patients might allow the administration of a higher dose intensity of chemotherapy than usual as well as the prolonged use of antiretroviral drugs. Novel clinicopathological entities In the past few years, new clinicopathological entities of HIV–NHL have been reported. Primary effusion lymphoma Primary effusion lymphoma (PEL) is characterized by HHV-8 infection of the tumour clone and by a peculiar tropism of the serous body cavities [6]. PEL is a rare disease, accounting for 5% of all AIDS–NHL. Clinically, PEL is characterized by growth in the liquid phase (i.e. ascitis, pleural or pericardial effusion) in the absence of detectable tumour masses [6,70–79]. Pathologically, however, PEL presents as multiple small tumour foci involving the serous membranes, which appear irregularly thickened without infiltrative or destructive growth patterns [70,79,80]. These clinical and gross pathological features indicate that PEL is a primary serous membrane neoplasm. Cytomorphological features of PEL cells bridge those of large cell immunoblastic and anaplastic large cell lymphomas [6,72,73,75,78], and the most common appearance is a mixture of immunoblast-like cells and other cell types, including anaplastic, multilobated or multinucleated large cells resembling the Reed–Sternberg cells of HD (Fig. 9). Although PEL exhibits an indeterminate immunophenotype [6,71–73,75], immunogenotypic studies have confirmed its B cell origin. The association of PEL cells with markers typical of the late stages of B cell differentiation, such as CD138/syndecan-1 and MET/hepatocyte growth factor co-expression, has been taken to suggest that PEL reflects a pre-plasmacell stage [81,82].Fig. 9.: Cytomorphology of primary effusion lymphoma. The tumour cells are characterized by nuclei that are irregularly shaped and variably chromatic with prominent, often multiple nucleoli. Multilobated or multiple nuclei are also seen. Cytospin preparation, haematoxylin-eosin stain, × 250.The molecular pathogenesis of PEL has been partly elucidated. Infection of the tumour clone by HHV-8 occurs in 100% of cases and is frequently, although not always, coupled to infection by EBV [71,72,77,83,84] (Table 1). The infection is characterized by a high viral load, approximating 60–100 copies of the HHV-8 genome per cell, and displays a marked restriction of viral gene expression, consistent with a pattern of latent infection. Although the precise pathogenetic contribution of HHV-8 to AIDS–PEL is still under investigation, several pieces of evidence suggest that HHV-8 may be required for AIDS–PEL pathogenesis. First, the association of HHV-8 with AIDS–PEL is consistent and specific among lymphoid neoplasms [6,70,71,77]. Second, AIDS–PEL express some HHV-8 genes, including vIL-6 and viral cyclin, that have the potential to affect B cell growth [85–89]. The HHV-8 viral cyclin is homologous to cellular D-type cyclins, a class of positive cell cycle regulators that are physiologically modulated by the p27Kip1 cell cycle inhibitor [85]. The expression of viral cyclin may account for the high proliferative index of AIDS–PEL, which is curiously accompanied by the sustained expression of the p27Kip1 cell cycle inhibitor [89]. The co-existence of p27Kip1 expression and the high proliferative index observed in PEL may be explained, at least partly, by the fact that HHV-8 v-cyclin is resistant to p27Kip1-modulated inhibition [90]. Despite the major pathogenetic role attributed to HHV-8 in the development of PEL, it is conceivable that the virus is not sufficient for the development of PEL because non-neoplastic B cells infected by HHV-8 are also detected at a certain frequency among HIV-infected individuals without lymphoma [91,92]. Furthermore, PEL is consistently monoclonal and harbours recurrent cytogenetic abnormalities, whereas lymphoproliferations driven solely by viral infection are generally polyclonal and devoid of clonal genetic lesions [93,94]. The differential diagnosis of PEL from other lymphomas involving the serous body cavities is not feasible on pure clinical grounds, but is facilitated by the biological characteristics of the tumour, such as positivity for HHV-8 infection and the absence of c-MYC alterations (Fig. 10). Conversely, cases of Burkitt's lymphoma initially presenting as lymphomatous effusion with no detectable solid mass are consistently negative for HHV-8 infection, whereas harbour molecular alterations of c-MYC (Fig. 10).Fig. 10.: Classification and differential diagnosis of non-Hodgkin's lymphomas involving the serous body cavities and presenting as lymphomatous effusion. The grid also emphasizes the clinical, morphological and biological heterogeneity of these lymphomas. AIDS-related lymphomas involving the serous body cavities are distinguished into primary and secondary lymphomatous effusions. Secondary lymphomatous effusions closely reflect the phenotype and genotype of the corresponding tissue-based lymphoma and are consistently devoid of human herpes virus 8 (HHV-8) infection. Primary lymphomatous effusions include primary effusion lymphoma (PEL) as well as rare cases of Burkitt's lymphoma (BL) or immunoblastic plasmacytoid lymphoma (IBPL). The distinction between PEL and other types of primary lymphomatous effusions is based on HHV-8 infection of the tumour clone. BL is also distinguishable on the basis of the consistent association with c-MYC translocation. ALCL, Anaplastic large cell lymphoma; EBV, Epstein–Barr virus; PAL, pyothorax-associated lymphoma.In contrast to other AIDS–NHL categories, which occur in both sexes and in all risk factor groups, PEL preferentially develops in homosexual men [6,70–72,77]. Also, in a given patient, PEL sometimes co-exists with other HHV-8-related diseases, namely KS and multicentric Castleman disease. At the Aviano Cancer Center, between 1987 and 1998 we observed eight cases of PEL accounting for 4% of all AIDS–NHL diagnosed in the same period of time [72]. In particular, we observed an association between PEL and multicentric Castleman disease and between PEL and KS. PEL presented exclusively as a malignant effusion in five patients and showed extension to visceral or distant sites in three patients. The CD4 cell count was low (< 50/mm3), suggesting that the disease is associated with a marked degree of immunodeficiency. With CHOP or CHOP-like regimens, the median survival was only 2 months. Plasmablastic lymphoma of the oral cavity AIDS-related plasmablastic lymphoma (AIDS–PBL) is associated with unique immunohistological features and a specific localization to the oral cavity mucosa [7]. Histologically, AIDS–PBL tumour cells display a blastic cell morphology with plasma cell differentiation (Fig. 11). This is consistent with the unusual immunophenotype of this lymphoma (CD20−, CD45−/+, CD79a−/+, EMA+, VS38c+, CD138/B-B4+), which reflects a marked plasma cell differentiation of the tumour clone. The molecular pathophysiology of AIDS–PBL is poorly understood. Infection by EBV occurs in approximately 60% of cases, although the expression of LMP-1 is restricted to a fraction of cases [7,95]. To date, no other molecular alterations have been detected in these lymphomas.Fig. 11.: Plasmablastic lymphoma of the oral cavity. The histological appearance of plasmablastic lymphoma (PBL) is relatively monomorphic. PBL displays a diffuse and cohesive growth pattern with large neoplastic cells exhibiting a squared-off appearance. These cells have a centrally or eccentrically placed nucleus with a single prominent nucleolus or several nucleoli. Haematoxylin-eosin stain, × 250.The incidence and distribution of AIDS–PBL are still unknown. In a consecutive series of 200 AIDS–NHL observed at the Aviano Cancer Center between 1984 and 1997, PBL accounted for 4.5% of cases and occurred in all risk groups. AIDS–PBL may develop in the presence of relatively sustained (> 100/mm3) CD4 cell counts and, accordingly, constitutes the first AIDS manifestation in a substantial fraction of cases. Patients usually present with limited disease, the most commonly involved sites being the gingiva and hard palate. Extraoral disease (with involvement of the gastrointestinal tract, bone marrow, lymph nodes) may occur, although it is rare. Although chemotherapy may lead to CR, the median survival is relatively short (6 months in our series). Perspectives Previously in this journal, Herndier and coworkers [96] reviewed the lymphomas in the HIV setting. In the past few years, however, an improvement of our knowledge of the molecular pathogenesis and clinical management of AIDS–NHL has occurred. Recent data on the molecular pathogenesis of systemic AIDS–NHL have led to the identification of new entities and of distinct and specific genetic pathways selectively associated with different types of AIDS–NHL. Taking into consideration the expression of LMP-1, CD138/syndecan-1 and BCL-6, a new histogenetic model for systemic AIDS–NHL and HD may be proposed (Fig. 12) [97]. Because most pathological categories of AIDS–NHL are also seen in non-immunosuppressed patients, the classification of these lymphomas may be based on the criteria adopted for lymphomas of immunocompetent hosts, and a separate classification system for AIDS–NHL is not recommended [98]. However, novel classification schemes of lymphoma should highlight the distinctive biological and clinical features of AIDS-related lymphomas compared with lymphomas of similar histology but arising in immunocompetent hosts [98].Fig. 12.: A proposed model for histogenesis of AIDS-related non-Hodgkin's lymphoma and AIDS-related Hodgkin's disease. The model is derived from the expression profile of BCL-6 and CD138/syndecan-1 (syn-1) throughout physiological B cell maturation. B cells within the germinal centre (GC) display the BCL-6+/syn-1− phenotype, whereas B cells that have exited the GC and have undergone further maturation towards the plasmacell stage exhibit the BCL-6−/syn-1+ phenotype. AIDS-related lymphomas displaying the BCL-6+/syn-1− phenotype, i.e. AIDS-related Burkitt's lymphoma (BL) and AIDS-related large non-cleaved cell lymphoma (LNCCL) (both systemic and primarily localized in the central nervous system), are postulated to originate from GC B cells. Conversely, AIDS-related lymphomas displaying the BCL-6−/syn-1+ phenotype, i.e. AIDS-related immunoblastic plasmacytoid lymphoma (IBPL) (both systemic and primarily localized in the central nervous system), AIDS-related Hodgkin's disease (HD) and AIDS-related primary effusion lymphoma (PEL), are postulated to derive from post-GC, pre-terminally differentiated B cells. In the case of AIDS-related lymphomas infected by Epstein–Barr virus (EBV), only the BCL-6−/syn-1+ phenotype is permissive for the expression of the EBV-encoded antigen LMP-1. Conversely, the expression of LMP-1 is consistently absent among AIDS-related lymphomas displaying the BCL-6+/syn-1− phenotype.The dramatic changes seen in the incidence of opportunistic infections and, consequently, in life expectancy since the advent of HAART encourage a more aggressive treatment of systemic AIDS–NHL. For example, high-dose chemotherapy with peripheral stem cell rescue could be tested, at least in low-risk patients with systemic AIDS–NHL, taking into consideration the adverse prognostic factors present in these patients. Moreover, some noteworthy data on the use of monoclonal antibodies for the treatment of NHL in the general population have been published [99]. In particular, rituximab, a chimeric murine/human monoclonal antibody that targets the CD20 antigen expressed on normal B cells and on more than 95% of B cell NHL, was initially tested in the setting of low-grade lymphomas. Similar trials are also ongoing in high-grade AIDS–NHL. In AIDS–HD, aggressive chemotherapy regimens such as Stanford V [100] or high-dose chemotherapy with peripheral stem cell rescue should be tested, taking into consideration the good results obtained in HIV-negative patients with the same unfavourable prognostic factors present in AIDS–HD. Finally, new clinicopathological subtypes are likely to emerge among both AIDS–NHL and AIDS–HD, as new molecular, virological and clinical findings are appreciated and with a possible distinct evolution and response to therapy. Acknowledgements The authors would like to thank Mrs Daniela Furlan for skillful assistance in the preparation of the manuscript.