- © 2000 by American Society of Clinical Oncology
Epstein-Barr Virus Infection Is Predictive of CNS Involvement in Systemic AIDS-Related Non-Hodgkin’s Lymphomas
- Antonella Cingolani,
- Roberta Gastaldi,
- Lucia Fassone,
- Francesco Pierconti,
- Maria Letizia Giancola,
- Maurizio Martini,
- Andrea De Luca,
- Adriana Ammassari,
- Carla Mazzone,
- Edoardo Pescarmona,
- Gianluca Gaidano,
- Luigi Maria Larocca and
- Andrea Antinori
- From the Departments of Infectious Diseases and Pathology, Catholic University; Hematology Department of Cellular Biotechnology and Hematology, La Sapienza University; National Institute for Infectious Diseases, Lazzaro Spallanzani, Istituto di Ricovero e Cura a Carattere Scientifico, Roma; Division of Internal Medicine, Department of Medical Sciences, Amedeo Avogadro University of Eastern Piedmont, Novara; Italy.
- Address reprint requests to Andrea Antinori, MD, National Institute for Infectious Diseases, Lazzaro Spallanzani, Istituto di Ricovero e Cura a Carattere, Via Portuense 292, 00149 Roma, Italy; email andrea.antinori{at}tin.it
Abstract
PURPOSE: This study aimed at correlating Epstein-Barr virus (EBV) infection of systemic AIDS-related non-Hodgkin lymphomas (AIDS-NHL) with the development of a CNS localization of the tumor.
PATIENTS AND METHODS: Demographic, epidemiologic, clinical, histologic, and virologic features were collected for all systemic AIDS-NHL patients included in the study (n = 50). Pathologic specimens were classified according to the working formulation for NHL and the Revised European-American Lymphoma classification. EBV infection in tumor tissue samples was studied by EBV small encoded RNA in situ hybridization; EBV-DNA detection in CSF was carried out by nested polymerase chain reaction using Epstein-Barr nuclear antigen-1–specific primers. In addition, selected EBV-positive lymphomas were subjected to a detailed characterization of EBV molecular heterogeneity.
RESULTS: Eleven patients had a CNS involvement at some point during their clinical history (four at diagnosis and seven at relapse). Thirty patients (11 with CNS involvement and 19 without) harbored EBV infection of the tumor. Sensitivity, specificity, and positive and negative predictive values of EBV-DNA detection in CSF for CNS involvement by lymphoma were 90%, 100%, 100%, and 97.6%, respectively. Factors significantly predictive of CNS involvement were EBV infection of the tumor (P = .003), an extranodal disease at diagnosis other than CNS (P = .006), and a non-CNS relapse (P = .01). In four cases of CNS involvement, EBV-DNA in CSF preceded any other sign of disease by a mean of 35 days.
CONCLUSION: These results show that EBV infection of the tumor clone significantly increases the risk of CNS involvement by systemic AIDS-NHL, without regard of specific molecular features. The detection of EBV-DNA in the CSF of AIDS-NHL patients may select cases with higher risk of CNS involvement and, therefore, may prove useful in the therapeutic stratification of these tumors.
HUMAN IMMUNODEFICIENCY virus (HIV)-infected patients are at high risk for the development of aggressive B-cell non-Hodgkin’s lymphomas (NHL).1-3 Although AIDS-related NHL (AIDS-NHL) have been observed since the beginning of the AIDS epidemic, the incidence of these disorders further increased during this decade with prolonged life expectancy of HIV-infected individuals.4 In contrast to other AIDS-related neoplasms, the incidence of systemic AIDS-NHL has been stable over time after the introduction of highly active antiretroviral therapies.5-7
Extranodal site localization is extremely frequent in systemic AIDS-NHL, occurring in 75% to 95% of cases. In particular, meningeal or parenchymal brain involvement has been variably reported as a main feature of extranodal disease,8-10 and meningeal relapse occurs in approximately 35% of treated cases.11 Serum lactate dehydrogenase (LDH), multiple extranodal sites, and bone marrow involvement have been identified as risk factors for CNS recurrence of NHL in the general population,12 whereas no data are available for HIV-infected patients.
Infection by Epstein-Barr virus (EBV) is common among AIDS-NHL and occurs with different rates according to the histologic features of the neoplasm, ranging from 30% in systemic AIDS-related Burkitt’s lymphoma, to 70% to 80% in systemic AIDS-related diffuse large-cell lymphoma, and to virtually all cases of AIDS-related primary CNS lymphoma (AIDS-PCNSL)13 and AIDS-related primary effusion lymphoma.14 The consistent association between EBV infection and AIDS-PCNSL allows a minimally invasive diagnosis of this tumor by detection of EBV-DNA in CSF using polymerase chain reaction (PCR).15
On this basis, the aims of this study were two-fold. First, we aimed at analyzing the correlation between EBV infection of the tumor clone and CNS involvement by systemic AIDS-NHL. Second, we evaluated the role of EBV-DNA in CSF as a diagnostic tool of CNS localization of systemic AIDS-NHL. Our results show that EBV infection is an independent risk factor for CNS involvement by AIDS-NHL, regardless of the genotypic features of the virus analyzed. In addition, detection of EBV-DNA in the CSF of systemic AIDS-NHL patients may facilitate the identification of cases with subclinical lymphoma localizations in the CNS and/or meninges that are not otherwise detectable by conventional diagnostic strategies.
PATIENTS AND METHODS
Patients
The clinical setting of the study was represented by the Department of Infectious Diseases of the Catholic University, Rome, Italy, and by the Hematology Section of the Department of Cellular Biotechnology and Hematology, University La Sapienza, Rome, Italy. The study enrolled all systemic AIDS-NHL patients who were observed at both institutions during a 4-year period for whom a pretreatment CSF sample was available. Demographic, epidemiologic, clinical, immunologic, histologic, and virologic features were collected for all patients. Clinical staging was uniformly performed by complete history and physical examination, bone marrow aspirate and biopsy, computed tomography of the chest, abdomen, and pelvis, computed tomography or magnetic resonance imaging of the brain, and lumbar puncture with cytologic and routine evaluation of the CSF. For staging purposes, the Ann Arbor Conference classification system was used. Patients were treated according to standard clinical protocols for systemic AIDS-NHL.16 All patients underwent prophylaxis of CNS localization with intrathecal administration of methotrexate. Neurologic involvement by the neoplasm was definitively diagnosed by the presence of lymphomatous cells in CSF or by histology.
Histopathologic Analysis and In Situ Hybridization
Pathologic specimens were histologically classified into AIDS-related diffuse large-cell lymphoma (AIDS-DLCL) and AIDS-related small non–cleaved-cell lymphoma (AIDS-SNCCL), according to the Revised European-American Lymphoma classification of lymphoid neoplasms.17 As previously reported,13 all neoplasms classified as AIDS-DLCL were further distinguished into two histologic subtypes, ie, large non–cleaved-cell lymphoma (LNCCL) and immunoblastic plasmacytoid lymphoma (IBPL), whereas all SNCCL were further subdivided into Burkitt’s lymphoma and Burkitt-like lymphoma.17 Moreover, other NHL subtypes, such as lymphoma with mixed morphology, or particular NHL entities, such as diffuse anaplastic large-cell lymphoma and plasmablastic lymphoma of oral cavity, were also identified. In situ hybridization (ISH) analysis of EBV-encoded small RNAs was carried out on paired paraffin tissue sections by using a cocktail of fluorescein-isothiocyanate–labeled oligonucleotides complementary to the two nuclear EBV small encoded RNAs (EBERs) (type 1/2), as previously reported.13 Samples of CSF for biomolecular analysis were collected for each patient before starting antineoplastic therapy, and EBV-DNA was amplified by a nested PCR of the Epstein-Barr nuclear antigen (EBNA)-1 region, as previously described.15
Analysis of EBV Molecular Heterogeneity
Definition of EBV type 1/2 was performed by PCR using primers specific for the EBNA-2 and EBNA-3C genes, as previously reported.18-20 Heterogeneity of the LMP-1 gene was investigated by PCR analysis using primers that explore the carboxyterminus of LMP-118,20 and allow the distinction between wild-type LMP-1 and the LMP-1 deletion variants known as Δ30LMP-1 and Δ69LMP-1. The distribution of EBNA-1 variants was determined by DNA direct sequencing of PCR-amplified products, as previously reported.20,21 DNA samples were amplified by a single-step PCR protocol using primers that amplify a 330-base pair fragment, containing most substitutions observed in EBNA-1.20,21 All sequences obtained were compared with the DNA sequence of B95.8, which represents the EBV prototype.22
Statistics
For statistical analysis, the association of CNS involvement with discrete variables was analyzed by contingency tables and Fisher’s exact test. The independent value of the association was controlled in a multivariate model by stepwise logistic regression analysis, using as covariates associations with a P value at entry more than .10. All P values were two-tailed. Sensitivity, specificity, and predictive values of EBV-DNA in CSF for diagnosis of CNS involvement were calculated using standard formulas.
RESULTS
Clinical Features of Systemic AIDS-NHL
During the study period (January 1995 to December 1998), 55 newly diagnosed systemic AIDS-NHL patients were enrolled. Five patients were excluded from the study because their tumor samples were not available for EBERs ISH analysis. Among the 50 eligible patients, 23 (46%) were female and 27 (54%) were males; the median age was 35.5 years (quartile [Q]1 to Q3, 32 to 42 years). Nineteen patients (38%) were intravenous drug users, whereas 31 (62%) had their HIV infection transmitted by a sexual route (10 men had homosexual encounters and 21 patients reported heterosexual contacts). Twelve patients (24%) had a previous AIDS-defining event. At diagnosis, the median CD4 cell count was 106/μL (Q1 to Q3, 40 to 216/μL), the median serum LDH level was 740 IU/L (Q1 to Q3, 400 to 1202 IU/L), 32 patients (64%) had B symptoms, 32 patients (64%) had extranodal lymphoma localization, and 34 patients (68%) were in Ann Arbor stage III-IV.
According to the Revised European-American Lymphoma classification, DLCL was diagnosed in 39 patients (78%), Burkitt’s lymphoma in eight patients (16%), and Burkitt-like lymphoma in three patients (6%). According to the Working Formulation, the systemic AIDS-NHL included in this study were classified as IBPL (n = 12; 24%), LNCCL (n = 18; 36%), and SNCCL (n = 11; 22%). A mixed morphology IBPL/LNCCL was displayed in five patients (10%), whereas anaplastic large-cell lymphoma and plasmablastic lymphoma of the oral cavity were found in one (2%) and three (6%) patients, respectively. Fourteen patients (28%) had a relapse that, in seven patients (50%), involved the CNS.
An extranodal disease, not including CNS involvement, was observed in 32 (64%) of the 50 patients at NHL diagnosis and in eight patients (16%) at relapse. At the time of lymphoma diagnosis, four patients (8%) had CNS involvement (three patients with definitive diagnosis and one patient with presumptive diagnosis), whereas, at relapse, seven patients (14%) had CNS involvement (six definitive diagnoses and one presumptive diagnosis). Both presumptive diagnosis (patients no. 5 and 11) were confirmed at autopsy as NHL localizations (Table 2).
Correlation Between EBV Infection of the Tumor Clone and Clinical Features of Systemic AIDS-NHL
EBERs ISH studies demonstrated EBV infection of the tumor clone in 30 patients (60%) with systemic AIDS-NHL (11 patients with CNS involvement and 19 without). EBV-DNA was detected in the CSF of nine out of 10 assessable patients with CNS localization of the disease and in none of the patients without CNS involvement. Sensitivity, specificity, and positive and negative predictive values of EBV-DNA detection for CNS involvement by lymphoma were 90% (95% confidence interval [CI], 54.1 to 99.5), 100% (95% CI, 89.1 to 100.0), 100% (95% CI, 62.9 to 100.0), and 97.6% (95% CI, 85.9 to 99.9), respectively.
Among all epidemiologic, clinical, immunologic, and histopathologic features investigated, the only factors predictive of CNS involvement were EBV infection of the tumor clone, extranodal disease at diagnosis, and a non-CNS relapse (Table 1). At logistic regression, the risk of CNS involvement was 20.11 (95% CI, 1.41 to 286.50; P = .003) for EBV infection of tumor tissue, 18.4 (95% CI, 1.20 to 280.90; P = .006) for having extranodal disease, and 13.7 (95% CI, 1.16 to 162.37; P = .01) for a relapse in a non-CNS site. No particular extranodal site was found to be specifically predictive of CNS involvement. None of the other demographic, epidemiologic, and clinical features, including age, sex, HIV transmission route, CD4 level, previous AIDS diagnosis, or neoplasia-related variables, such as serum LDH, presence of B symptoms, or stage, showed a significant association with EBV infection of the tumor clone. Analyses of the correlation between EBV infection and tumor morphology according to the Working Formulation showed that the probability was 1.80 (95% CI, 0.39 to 8.70; P = .39) for IBPL, 0.33 (95% CI, 0.08 to 1.28; P = .07) for LNCCL, and 1.02 (95% CI, 0.22 to 4.80; P = .90) for SNCCL. Cumulating IBPL with mixed IBPL/LNCCL, a probability of 2.55 (95% CI, 0.63 to 10.75; P = .13) was found.
Epidemiologic and Virologic Features According to CNS Involvement in 50 Eligible Patients With Systemic AIDS-NHL
Clinical and Virologic Features of Systemic AIDS-NHL With CNS Involvement
The detailed features of the 11 systemic AIDS-NHL with a CNS involvement are listed in Table 2. All but three patients developed a clinical neurologic syndrome at the time of CNS involvement. This was characterized by focal signs in five cases (45%) and by abnormal level of consciousness in six cases (54%). Four patients with lymphomatous cells in CSF never showed any abnormality at neuroradiologic examination during the course of the neoplasm. All patients with CNS involvement displayed EBV infection in their tumor tissue, and all but one patient had EBV-DNA detectable in the CSF. In four cases, the detection of EBV-DNA in CSF preceded the onset of clinical features and neuroradiologic findings by a mean of 35 days.
Descriptive Analysis of Systemic AIDS-NHL Patients With CNS Involvement
Molecular Features of EBV Infection in Systemic AIDS-NHL
The molecular features of EBV infection in systemic AIDS-NHL are summarized in Table 3. Overall, the data show that CNS localization of systemic AIDS-NHL occurs irrespective of the specific molecular features of the EBV strain associated with the lymphoma.
Molecular Analysis of EBV According to CNS Involvement
DISCUSSION
The results of this study unequivocally document that CNS involvement by systemic AIDS-NHL is strictly associated with EBV infection of the tumor clone. Also, EBV-DNA detection in CSF of patients with systemic AIDS-NHL is a highly predictive marker for diagnosis of CNS localization, sometimes preceding neuroradiologic findings of intracerebral tumor localization. Overall, these data may prove useful in refining the staging procedures and therapeutic stratification of systemic AIDS-NHL
The association between EBV infection and CNS involvement by systemic AIDS-NHL has relevant implications for the clinical management of these tumors. In fact, our results show that the risk of CNS localization in EBV-positive systemic AIDS-NHL is about 10-fold higher than the risk of EBV-negative cases. In addition to EBV infection, our data demonstrate that extranodal disease also predicts for CNS localization, as previously observed for NHL occurring in the general population.12 Consequently, our data suggest that CNS prophylaxis of systemic AIDS-NHL should be reserved only to those patients who have EBV infection of the tumor tissue, especially when an extranodal site of disease is documented at tumor staging. Conversely, patients with negative EBV transcripts in tumor tissue and an exclusively nodal disease might be excluded from routine prophylaxis against meningeal localization of systemic AIDS-NHL.
Based on the results presented in this study, EBV-DNA detection is a potentially useful tool for the rapid diagnosis of meningeal or parenchymal brain involvement in the context of systemic AIDS-NHL. In particular, 36% of cases harbored EBV-DNA in CSF before any other signs of CNS involvement, and in 27% of patients, EBV-DNA detection preceded the appearance of clinical features. The observation that positive EBV-DNA detection in the CSF precedes neuroradiologic and clinical signs of CNS localization of systemic AIDS-NHL is consistent with data derived from other AIDS-NHL involving the CNS, namely AIDS-PCNSL.23 Notably, predictive values of the EBV-DNA PCR test performed for systemic AIDS-NHL are very close to those obtained in AIDS-PCNSL, for which this technique has been validated for minimally invasive diagnosis15 and for treatment monitoring.24 Therefore, amplification of EBV-DNA sequences in CSF may represent a reliable procedure for the early detection of CNS localization of systemic AIDS-NHL and may allow for a prompt start of specific treatment of these neoplasms.
The causal relationship between EBV infection and CNS involvement by systemic AIDS-NHL needs to be further elucidated. This association has been found to be independent of other factors predictive of neurologic involvement, such as extranodal disease or relapse, or factors linked to the morphologic heterogeneity of these disorders. The fact that CNS involvement is strictly related to EBV infection of the tumor clone suggests an analogy with AIDS-PCNSL,13,25 in which EBV is a major pathogenetic element. However, systemic AIDS-NHL with CNS involvement differ substantially from AIDS-PCNSL because of the relatively higher CD4+ level (45% of patients with more than 100 cells/μL) and the heterogeneity of tumor morphology, which includes both DLCL and SNCCL, whereas virtually all AIDS-PCNSL are represented by DLCL. As already demonstrated for AIDS-PCNSL,26 no preferential occurrence of EBV type 1 or 2 has been observed in systemic AIDS-NHL with CNS involvement. Also, Δ69LMP-1 deletion variant, which has been found to be relatively frequent among AIDS-PCNSL,26 did not discriminate between systemic AIDS-NHL with or without a CNS involvement.
Independent of pathogenetic implications, this study suggests that the assessment of EBV infection in the tumor tissue and the analysis of EBV-DNA in CSF of patients with systemic AIDS-NHL may allow the identification of patients with an active CNS localization or patients who are at high risk of its development. On this basis, these assays should be used to select patients to early treatment or adequate CSF prophylaxis. Further studies are necessary to clarify the mechanisms by which EBV can drive clonal expansion in the CNS compartment.
Acknowledgments
Supported by the Istituto Superiore di Sanità, I and II Programma Nazionale di Ricerca sull’AIDS, 1997–1999; Ricerca Corrente e Finalizzata degli IRCCS.
Footnotes
-
Preliminary results were presented at the Third National AIDS Malignancy Conference, Bethesda, MD, May 26-28, 1999.
- Received January 4, 2000.
- Accepted May 30, 2000.
