HIV-1-associated lymphoma was first incorporated into the U.S. Centers for Disease Control and Prevention's (CDC) case definition of AIDS in 1985.(1,2) Prior to the use of effective antiretroviral therapy, HIV-1-associated lymphomas constituted 3 to 4% of all AIDS-defining illnesses reported to the CDC. Following the advent of effective prophylactic regimens for common HIV-1-associated opportunistic infections (eg, Pneumocystis jiroveci, Toxoplasma gondii, cytomegalovirus) and the development of antiretroviral monotherapies, the incidence of HIV-1-associated lymphomas remained constant at 1.6% per year.(3) In 1992, HIV-1-associated lymphomas were estimated to comprise 8-27% of approximately 36,000 newly diagnosed cases of lymphoma.(4) Without effective antiretroviral therapy, it is estimated that 5-10% of all HIV-infected individuals will have lymphoma as either an initial or subsequent AIDS-defining condition.(5) HIV-1-associated lymphomas thus represent a significant clinical entity within the spectrum of HIV-1-associated illnesses.

The advent of antiretroviral therapy initially failed to demonstrate a dramatic decrease in HIV-associated lymphomas comparable with that observed in opportunistic infections and Kaposi sarcoma. There is increasing evidence, however, that potent antiretroviral therapy does decrease the incidence of HIV-1-associated lymphomas,(6-10) and that antiretroviral therapy combined with antineoplastic chemotherapy (versus chemotherapy alone), improves outcome and survival.(6)

Approximately 95% of HIV-1-associated lymphomas are considered to be of B-cell origin because of cell surface expression of B-cell antigens (eg, CD19, CD20), evidence of immunoglobulin (Ig) gene rearrangement(s), and/or Ig production by tumor cells. HIV-1-associated lymphomas can be divided into at least two distinct clinical categories based on site of disease presentation; these two categories are in turn correlated with the extent of host immune function and response to chemotherapy. A 1991 review(11) of 2,500 cases of HIV-1-associated lymphomas revealed that approximately 80% arose in the periphery (ie, "systemic" lymphomas), and 20% arose in the central nervous system (CNS).(12-14) This distribution of lymphomas remains the same in the era of highly active antiretroviral therapy (HAART).(15)

For HIV-1-associated systemic lymphomas, widespread disease involving extranodal sites is common.(16-18) Symptoms at the time of presentation can be quite variable; 75% of patients will have peripheral CD4 T-lymphocyte counts >50/mm³ at presentation, and many patients will not have had prior opportunistic infections.(19) Approximately 75-80% of these lymphomas are classified histologically as large-cell lymphomas and the remaining 20-25% as Burkitt (or small-cell). Large cell lymphomas were historically categorized as large cleaved-cell, sclerosing variant large-cell, large-cell immunoblastic, plasmacytoid, immunoblastic clear-cell, polymorphous, immunoblastic, and immunoblastic with epithelial cell component. In an effort to standardize the nomenclature, improve diagnostic accuracy, and promote international consensus, the World Health Organization (WHO) has developed a different classification of lymphomas.(20) In particular, the more commonly observed HIV-associated large-cell lymphomas are now classified as either diffuse large B-cell lymphoma/centroblastic or diffuse large B-cell lymphoma/immunoblastic. Similarly, and in agreement with the WHO recommendations, HIV-associated Burkitt lymphoma is now categorized as classical Burkitt lymphoma, Burkitt lymphoma with plasmacytoid differentiation, or atypical Burkitt lymphoma.

HIV-1-associated primary CNS lymphomas usually arise de novo in patients with advanced AIDS; 75% of patients will have peripheral CD4 T-lymphocyte counts <50/mm³, and most patients will have had prior opportunistic infections.(2) Prognosis is uniformly poor. Symptoms are often indistinguishable from those of CNS toxoplasmosis, an opportunistic infection that occurs at a similar frequency, and diagnosis has been dependent on microscopic examination of tissue obtained by invasive brain biopsy. There has been recent interest in the use of target amplification methods (eg, polymerase chain reaction [PCR]) for detecting Epstein-Barr virus (EBV) in cerebrospinal fluid (CSF) as a surrogate diagnostic test for CNS lymphoma. One study involving 122 HIV-infected patients--42 with primary CNS lymphoma and 80 with nonmalignant conditions--using a nonstandardized nested PCR method determined that EBV PCR on CSF had a sensitivity of 80% (95% confidence interval [CI] 61-92%) and a specificity of 100% (95% CI 93-100%).(21) Brain thallium-201 single photon emission computerized tomography (Tl-201 SPECT) of 38 patients had a sensitivity and specificity of 86% and 83%, respectively, for diagnosing lymphoma.(22) Studies are underway to assess the diagnostic yield of EBV PCR on CSF combined with Tl-201 SPECT for CNS lymphoma.

Early in the AIDS epidemic, HIV-1-associated lymphomas were examined for evidence of clonality and presence of cofactors traditionally associated with B-cell malignancies (eg, infection with EBV), which is highly associated with endemic Burkitt lymphoma and B-cell lymphoproliferative disease in immunosuppressed patients, and c-myc translocations, which are highly associated with sporadically occurring Burkitt lymphomas. The molecular features of primary CNS lymphomas differed from those of systemic lymphomas. The majority of primary CNS lymphomas were large-cell lymphomas that were virtually all monoclonal tumors monoclonally infected with EBV and lacking c-myc rearrangement.(23,24) Systemic lymphomas, in contrast, were heterogeneous at the molecular level; both monoclonal and polyclonal tumors were observed (the latter defined by the inability to detect monoclonal Ig gene rearrangement in at least 5% of cells comprising the tumor).(14) EBV infection was not universal (and if detected, was not correlated with clonality), and c-myc translocations were present in only a subset of those tumors that were monoclonal. Genetic lesions typically associated with other malignant processes (eg, p53 and/or ras mutations observed in only a subset of lymphomas, and mutations in the retinoblastoma gene [RB1](25)) have not been consistently detected in systemic HIV-1-associated lymphomas. These two clinically and molecularly distinct lymphoma classes--CNS and systemic--are still relevant in the HAART era, although CNS lymphoma has dramatically declined in the setting of HAART.

The discovery of "polyclonal" lymphomas lacking EBV or c-myc translocations was initially controversial but later substantiated by multiple groups.(26,27) An autopsy study of three patients who died of widespread lymphoma demonstrated polyclonal lymphoma occurring as metastatic lesions in the liver, lung, and other organs, and provided evidence that polyclonal lymphomas were in fact as aggressive as, and clinically indistinguishable from, previously reported monoclonal lymphomas.(28) In a 1995 clinical outcome study, polyclonal lymphoma patients survived significantly longer than patients with monoclonal lymphomas, adding prognostic significance to clonality measurement.(29)

The observation that HIV-1-associated primary CNS lymphomas occur in individuals with advanced AIDS suggested a pathogenetic model wherein clonal outgrowths of B cells proliferate in response to an opportunistic infection with EBV in the absence of immune regulation. EBV has been proposed to play a pathogenetic role in EBV-associated B-cell lymphomas in the setting of immunodeficiency (eg, iatrogenically immunosuppressed allograft recipients), presumably by conferring a selective growth advantage to a clone of cells infected with EBV. The pathogenetic mechanism may be similar to that by which EBV immortalizes B cells in vitro and may explain monoclonal EBV-associated B-cell lymphomas arising in all mice with severe combined immunodeficiency disease (SCID) engrafted with peripheral blood lymphocytes obtained from EBV-seropositive donors.(30,31) In the setting of effective antiretroviral therapy, the frequency of opportunistic infections declines dramatically.(6) The recent dramatic decline in primary CNS lymphoma in parallel gives further evidence that this disease may be an EBV opportunistic infection.(32)

In contrast, the heterogeneous molecular characteristics of HIV-1-associated systemic lymphomas could not be so easily explained, leading to the development of a multifactorial pathogenetic model. Early molecular analyses of the immunoglobulins produced by HIV-1-associated systemic lymphomas provided strong support for this model.(33-38) In this model, HIV-1-associated lymphomas are part of a continuum beginning with polyclonal proliferation of B cells and culminating with the outgrowth of a transformed B-cell clone. Polyclonal proliferation of B cells occurs as the initial step because of inadequate B-cell control from T cells depleted or rendered dysfunctional by HIV-1 infection and subsequent host immunodeficiency. Continuously proliferating B cells would provide a population of B cells at increased risk for accumulation of additional as yet undefined genetic events that would ultimately lead to the malignant transformation and outgrowth of B-cell clones. For those HIV-1-associated lymphomas that have been observed clinically, polyclonal lymphomas would represent events occurring during the early and middle stages of this model, and monoclonal lymphomas the final stage.

More recent studies have provided further evidence to support this model. Grulich and colleagues identified B-cell stimulation and prolonged immunodeficiency as risk factors for lymphoma development in HIV-1-infected individuals.(22) Analysis of Ig variable (V) gene use from circulating B cells of HIV-1-infected individuals demonstrated aberrant and unstable expression of Ig genes, providing strong evidence for a dysregulated humoral immune system in HIV-1 infection.(39) HIV-1 antigens, however, are most likely not the major antigens driving B-cell proliferation and placing them at risk for additional genetic events leading to lymphoma. Although early studies demonstrated that a few AIDS lymphomas produced Igs directed against HIV-1 antigens, the bulk of such studies have eliminated HIV-1 antigens as the lymphoma Ig cognate antigen and have failed to identify possible cognate antigens.(33,34,40,41)

At the same time that polyclonal lymphomas were identified, there was an intense effort to determine the role of HIV in lymphomagenesis. Many early studies failed to demonstrate the presence of HIV in tumor-associated B cells. Although most AIDS-associated lymphomas are of B-cell origin, an unusual monoclonal AIDS-associated T-cell lymphoma composed of CD4⁺ T cells and containing a single copy of integrated HIV was identified.(42) The HIV integration site in this clonal tumor was determined to be in the 3´ exon of the fur gene, just upstream of the c-fes oncogene. Subsequent molecular studies showed that the long terminal repeat (LTR) of the integrated HIV acted as an enhancer element for c-fes, resulting in an upregulation of Fes expression.(43) These data suggested that HIV, through an insertional mutagenesis process, could cause cellular transformation, at least in this particular T-cell lymphoma.

Detailed molecular studies from four patients with HIV-positive macrophage-rich tumors revealed monoclonally integrated HIV within tumor biopsies.(44) Monoclonally integrated HIV could not be detected, however, in tissue displaying hyperplastic (nontumor) lymphoproliferations at different sites in three of the four patients. The site(s) of HIV integration in each patient were determined with inverse PCR (iPCR), a technique that permits rapid identification of clonal integration sites. The region in which HIV was integrated was similar in all three non-T-cell lymphoma cases, just upstream to the c-fes oncogene. The clonally integrated HIV was demonstrated to be present exclusively in the macrophages within the tumors.(45)

HIV-related insertional mutagenesis, however, does not readily explain the observation that most HIV-associated lymphomas are of B-cell origin and lack evidence of HIV infection. In one model, macrophages harboring HIV integrated at a site that would enhance clonal macrophage proliferation (Fes mediates macrophage activation by M-CSF, GM-CSF, and IL-3) could result in a secondary proliferation of surrounding lymphocytes, which would then be susceptible to additional events (eg, EBV infection, c-myc translocation) leading to transformation.(45) This model is consistent with that proposed for Hodgkin disease wherein the Reed-Sternberg cell, which is the acknowledged monoclonal malignant cell, presumably provides growth factors for secondary proliferation of adjacent lymphocytes and fibroblasts that are in turn associated with Hodgkin disease.(46)

A subset of lymphomas occurring in HIV-1-infected individuals do not fit into the previously discussed categories. These unusual lymphomas have unique clinical or molecular features warranting discussion, even though their pathogenesis is poorly understood.

"Body Cavity-Based" HIV-1-Associated Lymphomas (BCBL), or Primary Effusion Lymphoma (PEL)

An unusual type of HIV-1-associated lymphoma is associated with human herpes virus, type 8 (HHV-8), a herpesvirus initially described in association with Kaposi sarcoma.(47) AIDS-related lymphomas reported to be associated with HHV-8 are aggressive fluid-phase lymphomas and have been variously termed "body cavity-based lymphoma", "AIDS-related lymphomatous effusion", "CD30 (Ki-1)-positive anaplastic large-cell lymphoma", and "primary effusion lymphoma".(48-55) These lymphomas arise in various body cavities (ie, pleural, peritoneal, pericardial) in the absence of lymph node or spleen involvement, although postmortem examination demonstrates widespread infiltration of adjacent serosal surfaces by the malignant cells.(55) These lymphomas typically occur in individuals with end-stage HIV-1-related disease, and outcomes have been uniformly poor. Little is known, however, regarding the effect of HAART on prognosis.(55) The malignant cells of these lymphomas have an unusual immunophenotype and typically display CD38 only (a subset coexpress B-cell markers CD19 and/or CD20). Although HHV-8 is universally present in these body cavity-based lymphomas, they are otherwise heterogeneous at a molecular level (ie, tumors can be monoclonal or polyclonal, and EBV infection or c-myc translocation is not universally present). Although multiple genetic abnormalities have been identified using comparative genomic hybridization analysis, many of the identified abnormalities are shared with other lymphomas, including non-HIV-1-associated lymphomas.(41) Growth of these tumors in vitro and in an animal xenotransplant model is mediated in part by autocrine production of the B-cell growth factor IL-6.(56) Induction of this growth factor and of vascular endothelial growth factor (VEGF), which inhibits apoptosis, appear to be driven in part by infection of lymphoma cells by HHV-8.(57) Induction by HHV-8 of macrophage inflammatory protein 1 (MIP-1) may also contribute to lymphomagenesis through attraction of other growth factor-producing cells into the lymphoma environment.(58) Other mechanisms by which HHV-8 seems to influence growth of these lymphomas include the production of viral homologues of known lymphoma growth factors, the best studied of which is viral IL-6, which shares growth-promoting activities with human IL-6.(59,60)

HIV-1-Associated Hodgkin Disease

Reports of 114 cases of HIV-1-associated Hodgkin disease have identified major differences in the clinical presentation and course of the disease in HIV-1-infected individuals compared with cases arising in immunocompetent individuals.(61-66) The age at presentation in HIV-1-infected individuals (mean age at presentation, 32 years, with a large group presenting between 35 and 49 years) does not follow the bimodal age distribution that has been observed for non-HIV-infected individuals. Furthermore, 50-89% of patients with HIV-1-associated Hodgkin disease, as compared with 40% of non-HIV-infected individuals, present with widespread disease (clinical stages III and IV), and involvement of bone marrow, CNS, gastrointestinal tract, and skin is common. A high proportion of HIV-1-associated Hodgkin disease is linked to EBV infection, implicating pathogenetic elements encoded by EBV.(67) The clinical course of HIV-1-associated Hodgkin disease is variable. In general, most HIV-1-infected patients respond clinically to conventional therapy but remain at high risk of mortality from opportunistic infections.(61-66)

HIV-1-Associated T-Cell Lymphomas

A few HIV-associated peripheral or cutaneous ("pseudo-Sezary syndrome") T-cell lymphomas have been reported. Prognosis is poor, and survival beyond 9 months is unusual. For one HIV-1-associated T-cell lymphoma that underwent extensive molecular analysis, HIV-1 was demonstrated to be clonally integrated just upstream to the c-fes oncogene within the genomic DNA from these tumor cells.(44,68) This finding was remarkable because it provided the first evidence that HIV-1 itself, through insertional mutagenesis, could be directly implicated in malignant transformation.

We described earlier studies demonstrating the presence of clonal macrophages harboring HIV integrated in a single site in polyclonal B-cell lymphoproliferative states or lymphomas, and propose that these clonal macrophages play a central initiating role in early lymphomagenesis wherein other immunologic factors play secondary and tertiary roles leading ultimately to lymphoma. Many molecular and virologic events occur that appear to be critical for lymphomagenesis; however, HIV might contribute directly to this process by driving early stages of lymphomagenesis through expansion of clonal macrophages and antigen-driven B-cell proliferation. Loss of B-cell maturational control has been demonstrated by studies of immunoglobulins associated with AIDS-related lymphoma. The randomly mutated nature of immunoglobulin variable region genes in lymphoma suggests that lymphomas may be outgrowths of B cells responding to any antigen. The overall immunodeficiency induced by HIV infection also plays a likely role, supported by recent and encouraging observations of decreased frequency of lymphomas in patients treated with HAART. Most AIDS-related lymphomas appear to be outgrowths of antigen-driven B cells with growth control influenced by abnormal T-cell and antigen-presenting cell processes. Further studies will be required to confirm or refute this general scheme.