Diagnosing Central Nervous System Lymphoma in the Setting of AIDS: a Step Forward

The accurate diagnosis of primary central nervous lymphoma (PCNSL) is one of the most vexing problems in the assessment of patients with human immunodeficiency virus (HIV) infection. Such patients generally present with focal central nervous system (CNS) lesions, and in this setting, PCNSL must be separated from a variety of other conditions that exhibit such lesions, including cerebral toxoplasmosis and other opportunistic infections. The frequency of PCNSL is more than 3000-fold higher in patients with acquired immunodeficiency syndrome (AIDS) than the general population, and about 20%–30% of CNS lesions in patients with AIDS are ultimately found to be PCNSL (1–3). The definitive diagnosis of PCNSL generally requires a brain biopsy (4). While generally safe, this procedure can be associated with serious sequelae and even death. Moreover, because the procedure involves craniotomy, it poses some risk of HIV infection for the neurosurgical team. [18F]Fluoro-2-deoxyglucose-positive emission tomography has some ability to discern between PCNSL and Toxoplasma abscesses (5), but it only discriminates partially between the two conditions and is available in only a few hospitals. For these reasons, in many medical centers it has become standard practice to treat AIDS patients empirically with anti-toxoplasmosis therapy if focal CNS lesions and anti-Toxoplasma antibodies are present, reserving biopsy for those patients who are seronegative for anti-Toxoplasma antibodies or fail to respond to treatment (6). However, this approach can lead to substantial delays in diagnosis. Even for those patients who are seronegative or fail to respond to anti-Toxoplasma therapy, there is often a reluctance to perform a brain biopsy for the reasons noted above, and many cases of AIDS-related PCNSLs are not diagnosed prior to death (7–9). For example, in one autopsy study, only one third of the patients with PCNSL were diagnosed antemortem (9). For these reasons, medical teams are periodically forced to consider administering empiric CNS radiation to patients with a presumptive diagnosis of PCNSL (10). The difficulties in diagnosing PCNSL (as well as other lymphomas) may have also confounded accurate epidemiologic studies of AIDS-related lymphomas. It is commonly cited that the proportion of HIV-infected patients who develop non- Hodgkin's lymphoma ranges from 2% to 10%. However, reports of autopsy studies of patients with AIDS have detected higher proportions, on the order of 20% (9), and a cumulative incidence of approximately 30% has been observed in at least one cohort of patients with AIDS who were followed carefully for several years (11). As noted above, a substantial number of cases of CNS lymphoma may go undetected antemortem, leading to an underestimation of the role of this lymphoma in the morbidity and mortality caused by HIV infection. The accompanying article by Cingolani et al. (12) shows that detection of cerebrospinal fluid (CSF) Epstein-Barr virus (EBV)-DNA by polymerase chain reaction (PCR) in HIV-infected patients is reliably associated with PCNSLs, lending support to the use of this assay as a diagnostic test for this condition. The basis for this approach is the observation that nearly all HIV-associated PCNSLs (as well as a number of systemic lymphomas in HIV-infected patients) are associated with EBV infection (13). Also, the development of lymphomas in HIV-infected patients has previously been shown to be preceded by an increase in the number of EBV-infected cells in the patients' lymphoid tissues (14). In 1993, Cinque et al. (15) reported that the EBNA-1 gene of EBV-DNA was detected in the CNS of each of the 17 patients with PCNSL but in only one of 68 patients without lymphoma by use of nested PCR. Interestingly, in four patients, EBV-DNA was detectable in the CNS before lymphomatous lesions could be visualized by computerized tomography or magnetic resonance imaging. The accompanying paper provides evidence that this approach can be used to diagnose PCNSL prospectively in HIV-infected patients with brain lesions; none of the 61 patients without PCNSL had detectable EBV-DNA, whereas 24 (80%) of 30 of the patients with PCNSL had detectable EBV-DNA sequences. This study provides strong evidence that measurement of CSF EBV-DNA can be a useful tool in the diagnosis of AIDSrelated PCNSL. Given the limitations of the present, empiric approaches to treatment (in which patients with focal CNS lesions sometimes receive radiation therapy largely because of failure to respond to anti-Toxoplasmosis therapy), this assay has the potential of representing an important diagnostic advance. It is, in fact, quite remarkable that an assay of viral activity can have such diagnostic accuracy for a tumor. However, the use of such an indirect approach without confirmatory pathology must provoke some intrinsic discomfort. Furthermore, it is worth stressing that this technique has been studied in relatively few patients and that not all cases of AIDS-related brain lymphoma are detected. Indeed, Arribas et al. (8) have reported that only three of six patients with PCNSL could be detected by this approach. Also, the very sensitivity of PCR technology (particularly if nested) can lead to false-positive results from laboratory contamination with DNA. In this regard, it is worth echoing the warning of Cingolani et al. that, while this assay may have a low false-positive rate in patients with HIV infection, this rate may climb to unacceptable levels in other patient populations in which the incidence of PCNSL is lower (perhaps even among HIV-infected patients with higher CD4+ cell counts) or in other conditions, such as primary EBV infection, that may involve substantial EBV replication. Thus, while the assay has the potential of substantially improving the ability to diagnose PCSNL, additional studies are warranted at this time to confirm the results of the Italian study and to examine the use of this assay in other settings. In the future, it may also be worth exploring whether PCR technology can be used to detect clonal rearrangements of immunoglobulin heavy-chain genes in CSF lymphocytes. This approach combined with detection of EBV-DNA might increase the positive predictive value of CSF testing for the diagnosis of PCNSL. The high incidence of certain unusual tumors in specific populations was one of the first indicators of the AIDS epidemic, and these opportunistic tumors continue to be a cause of substantial morbidity and mortality in HIV-infected individuals. The ability to identify patients with AIDS-associated PCNSL by assaying their CSF for EBV-DNA underscores our emerging appreciation of the importance of viruses in the pathogenesis of AIDS-related tumors. Indeed, with the recent discovery of Kaposi's sarcoma-associated herpesvirus and the evidence that it is an important factor in the pathogenesis of Kaposi's sarcoma and body-cavity lymphoma (16,17), it appears that most AIDSassociated tumors have a viral etiology. Research directed at understanding these relationships has the potential of leading to pathogenesis-based therapies or, as exemplified by the accompanying paper, to diagnostic tests that identify patients who either have the tumors or who are at high risk of developing them. It is quite possible that the impact of AIDS-associated PCNSL will increase as HIV-infected patients live longer because of treatment with highly active antiretroviral drugs and as we improve our ability to prevent and treat opportunistic infections. This assay may allow the diagnosis of cases of PCNSL that are now missed and thus help reveal the true incidence of this tumor. The present difficulty in diagnosing AIDS-associated PCNSL also poses an impediment to therapeutic trials. By enabling accurate early diagnosis, this assay may also help the clinical testing of new therapies for this condition. Finally, it will be of interest to see if this assay has value in monitoring the response of patients with PCNSL to therapy.