Primary Central Nervous System Lymphoma

Data Sources and Extraction

Using a MEDLINE search, we sought to identify all papers published in the English language between January 1980 and September 1992 that dealt with lymphoma and the brain, spinal cord, cerebral spinal fluid, or central nervous system. A search was also done of all literature dealing with AIDS, human immunodeficiency virus (HIV), lymphoma, and the central nervous system. Several series included overlapping patients. When these cases were identified, we extracted only the missing data for each case and attempted not to duplicate data. Patients were included in this analysis only if lymphoma was histologically documented.

Data Synthesis

Forty published series of predominantly non-AIDS-associated primary central nervous system lymphoma [16-53] and 32 series of AIDS-associated primary central nervous system lymphoma [54-85] were identified. Some data were present for 792 patients with non- AIDS-associated primary central nervous system lymphoma and 315 patients with AIDS-associated primary central nervous system lymphoma, although complete information regarding patient characteristics, type of treatment, and survival duration were seldom available for each person. The mean age of the patients without AIDS was 55.2 years; for the patients with AIDS it was 30.8 years (Table 1). The ratio of men to women was 1.35 for the patients without AIDS compared with 7.38 for the patients with AIDS. The average duration of symptoms before a definitive diagnosis was established was similar in the two groups (2.8 months for patients without AIDS compared with 1.8 months for patients with AIDS). A history of opportunistic infections or Kaposi sarcoma was present in 115 of 143 patients with AIDS (80%). Primary central nervous system lymphoma was the AIDS-defining illness in the remaining 28 patients (20%).

The types of initial symptoms differed only slightly when the two groups were compared (Table 1): Specific neurologic deficits, mental status changes (including behavioral changes), seizures, and symptoms and signs suggestive of increased intracranial pressure (headache, nausea, vomiting, papilledema) occurred in 56%, 35%, 11%, and 32% of all patients without AIDS, respectively, and in 51%, 53%, 27%, and 14% of patients with AIDS.

Radiographic imaging included 73 angiograms, all in patients with non-AIDS-associated primary central nervous system lymphoma. Sixty tumors were avascular, 10 showed a tumor blush, and 3 showed increased venous drainage (Table 2). Primary central nervous system lymphoma developed in supratentorial sites three times as often as in infratentorial locations. In addition, more than 60% of tumors were in a periventricular location, usually involving the corpus callosum, basal ganglia, or thalamus. Rarely, patients without obvious brain parenchymal tumors presented with only primary meningeal involvement. Findings from initial CT scans were similar in both groups of patients, with 80 of the 201 (90%) noncontrast scans demonstrating isodense or hyperdense lesions in the patients without AIDS compared with 61 of 167 (91%) scans in the patients with AIDS. Almost all lesions showed some degree of contrast enhancement (97% in patients without AIDS and 90% in patients with AIDS). Although the pattern of homogeneous enhancement (55% of patients without AIDS compared with 60.9% of patients with AIDS) was similar in the two groups, 52% of all AIDS-associated primary central nervous system lymphoma scans showed some degree of ring enhancement, a pattern not seen in the non-AIDS group. The other major radiographic difference between the two groups of patients was that multifocal disease was twice as common in patients with AIDS than in patients without AIDS (52% compared with 25%, respectively).

Diagnosis was achieved by surgical resection, rather than biopsy, in 62% of all patients without AIDS, whereas only 30% of patients with AIDS had such resections (Table 3). Spinal fluid contained malignant cells in 31% of patients with non-AIDS-associated primary central nervous system lymphoma. Only 13 instances of spinal fluid examination were reported in patients with AIDS and primary central nervous system lymphoma, of which 3 showed cells that were consistent with malignant lymphoma (23%) (Table 3).

When immunophenotyping was done, almost all lymphomas were of B-lymphocyte origin (Table 3). Differences in histologic pattern, using the Working Formulation terminology, showed only 22% of patients without AIDS having high-grade histologic findings (immunoblastic or small noncleaved cell) compared with 60% of patients with AIDS with the more aggressive, high-grade histologic findings. When reported, almost all AIDS-associated primary central nervous system lymphomas contained Epstein-Barr virus genomic DNA, whereas few tumors from immunocompetent patients contained Epstein-Barr viral DNA.

Treatment strategies were defined for 86% of the patients without AIDS, but only 55% of patients with AIDS had treatment decisions reported, possibly reflecting the relatively high number of cases in patients with AIDS that were diagnosed after death (Table 4). Radiation therapy represented the sole form of treatment in a similar percentage of patients in both groups (65.8% for patients without AIDS and 66.1% for patients with AIDS). Chemotherapy was more likely to be added to the treatment regimen in patients without AIDS (25%) compared with patients with AIDS (8%).

Without therapy both groups fared equally poorly (mean survival, 2.7 months for patients without AIDS; 0.9 months for patients with AIDS). However, the overall survival of all treated patients was substantially higher for patients without AIDS (18.9 months) compared with those with AIDS (2.6 months) (Table 4).

Epidemiology

Epidemiologic data strongly suggest that primary central nervous system lymphoma is increasing in incidence. Although the enhanced frequency in immunocompromised patients is related in part to the growing number of patients with iatrogenic immunosuppression (organ allograft recipients) and the longer survival of patients with congenital immunodeficiencies (owing to improved medical support), the principal reason for the large increase in case of primary central nervous system lymphoma in immunocompromised persons is the AIDS epidemic. A recent report from the National AIDS Cooperative Group described the development of eight cases of lymphoma in 55 patients on trials of antiretroviral drugs (azidothymidine), five of which were primary central nervous system lymphoma [85]. Although this figure is small, it does suggest a larger number of cases of primary central nervous system lymphoma than might otherwise be expected. Thus, as severely immunosuppressed patients survive longer, the incidence of this tumor may continue to rise. Less clear than the reasons for the growing frequency of this disease in immunodeficient patients is the explanation for the threefold increase in primary central nervous system lymphoma in immunocompetent persons during the last decade. Some observers have suggested that increased screening, with greater accessibility to MRI and CT technology, may partly account for the apparent increase in the incidence of this disease. This explanation, however, cannot account for the higher percentage of primary central nervous system lymphomas observed in resected or biopsied brain tumor specimens during the last 10 years. It therefore appears that the increase in the incidence of this disease is a true epidemiologic observation.

Biology

The cause of the neoplastic lymphocyte clone in non- AIDS-associated primary central nervous system lymphoma remains speculative. One possibility is that the neoplastic transformation occurs in one of the lymphocyte subpopulations that routinely inhabit the central nervous system. Alternatively, neoplastic transformation may occur in a systemic population of lymphocytes that already possess a specific tropism for the central nervous system (through expression of specific cell-surface adhesion molecules) or develop such a tropism after the transformation events. Studies comparing the immunophenotypic patterns of primary central nervous system lymphoma with those of systemic lymphomas, however, have not shown any substantial differences.

A third hypothesis for the cause of primary central nervous system lymphoma is based on the idea that the central nervous system is an immunologic sanctuary site. Neoplastic lymphocytes may be systemically eradicated by an intact immune system but find relative protection within the central nervous system. This theory could explain the curious observation that systemic dissemination of primary central nervous system lymphoma is rare, even in advanced stages of disease. Regardless of where the malignant cells originate, radiologic and pathologic data suggest that once within the central nervous system, these cells proliferate in the subarachnoid space or the substance of the choroid plexus or both and subsequently disseminate throughout the central nervous system [86].

The strikingly high incidence of primary central nervous system lymphoma in patients with AIDS is paralleled by a similarly high incidence of primary central nervous system lymphoma in other immunosuppressed patients, thus implicating the immune system in the pathogenesis of this tumor. Although many types of immunosuppression predispose to primary central nervous system lymphoma (congenital immunodeficiencies such as the Wiskott-Aldrich syndrome, the immunoglobulin-A deficiency syndrome, severe combined immunodeficiency syndrome, and idiopathic immunodeficiencies, such as those associated with systemic lupus erythematosus and rheumatoid arthritis), the best studied group of these patients are renal and cardiac allograft recipients. Penn [87] reported that the risk for developing lymphoma (half of which were primary central nervous system lymphomas) in a large group of renal allograft recipients was 350 times that of the general population. Weintraub and Wernke [88] reported three cases of primary central nervous system lymphoma in 182 heart recipients, a rate even higher than that of the kidney recipients. Several important similarities exist between primary central nervous system lymphoma in patients with AIDS and patients who are immunosuppressed. The tumor cells in these patients are generally oligoclonal or polyclonal, in contrast to the monoclonal tumors almost always found in the primary central nervous system lymphomas in immunocompetent patients. Another similarity is the predominance of high-grade histologic findings in immunosuppressed patients and patients with AIDS compared with the low- or intermediate-grade histologic findings that are frequently observed in immunocompetent patients. Although the data we accumulated concerning pathologic findings are somewhat difficult to interpret because of the several different histopathologic classifications used, it is clear that high-grade, immunoblastic, and small, noncleaved histologic findings appear in more than 60% of the primary central nervous system lymphomas in patients with AIDS but in only 20% of primary central nervous system lymphomas in patients without AIDS (Table 3).

A third similarity in patients with AIDS and organ allograft recipients is the tendency for the degree of immunosuppression to be related to the risk for developing primary central nervous system lymphoma. Conversely, primary central nervous system lymphomas seem to regress in association with a reduction in the dose of immunosuppressive drugs. An analogous situation appears to occur in patients with AIDS in that as the patient's CD4 lymphocyte count declines (and thus the patient becomes increasingly immunosuppressed), the risk for primary central nervous system lymphoma increases.

Possibly the most significant causative similarity in patients with AIDS and immunocompromised patients who have primary central nervous system lymphoma is the consistent finding of Epstein-Barr virus genomic DNA within the lymphoma cells [89-92], an observation that is uncommon in primary central nervous system lymphoma in immunocompetent persons [93]. This relation is intriguing in light of the ability of Epstein-Barr virus to induce B-lymphocyte proliferation in vitro and the strong epidemiologic evidence linking Epstein-Barr virus to endemic Burkitt lymphoma [94, 95]. Conceivably, Epstein-Barr virus may infect certain B-lymphocyte subpopulations, causing a clonal expansion of cells, an expansion that is probably limited by the regulatory mechanisms of an intact immune system. Such mechanisms probably include immune surveillance by natural killer cells and cytotoxic T lymphocytes as well as by T-lymphocyte-mediated B-cell growth regulation through the release of various lymphokines and cell-membrane-associated signals. In the primary central nervous system lymphoma-associated immunodeficient syndromes, however, the T-lymphocyte population is quantitatively (AIDS) or qualitatively (cyclosporine-treated) abnormal. This abnormality could allow Epstein-Barr virus-induced B-lymphocyte proliferation to proceed unrestrained, eventually leading to the formation of polyclonal tumors. Alternately, one specific clone of proliferating B cells may undergo a genetic alteration, giving it a selective growth advantage over the other B lymphocytes and resulting in a monoclonal tumor. The propensity of these lymphomas to grow in the brain may be explained by the decreased immune surveillance normally found within the central nervous system. Thus, primary central nervous system lymphoma in the immunocompromised patient may be the result of a process of uncontrolled Epstein-Barr virus-induced B-lymphocyte proliferation, and possibly subsequent neoplastic transformation, in an immunologically privileged site in a patient with baseline impaired immune surveillance.

Although this model may be generally correct, it is clearly oversimplified. That not all immunodeficient states predispose to primary central nervous system lymphoma suggests that subtle imbalances must be operative in its pathogenesis. A fuller understanding of this disease awaits insights into the molecular biology of lymphocyte transformation and the interactions of immune effector cells with the central nervous system.

Clinical Presentation

In addition to differences in the prevalence of Epstein-Barr virus genomic DNA and histologic findings between AIDS-associated and non-AIDS-associated primary central nervous system lymphoma, patient characteristics differ substantially in these two groups. Not surprisingly, patients with AIDS and primary central nervous system lymphoma are significantly younger (31 compared with 55 years) and comprise a higher percentage of men than the non-AIDS group. Whereas the average duration of symptoms before diagnosis is similar in the two groups, the range of symptom duration in the non-AIDS group is broad, indicating that some patients have symptoms for many months (possibly years) before they are diagnosed. This finding reflects the heterogeneity of this disease in immunocompetent persons. Such is not the case for primary central nervous system lymphoma in patients with AIDS, where the duration of symptoms rarely exceeds 2 to 3 months, reflecting a more homogeneously aggressive course.

Patients with AIDS-associated primary central nervous system lymphoma tend to have differing initial symptoms than other patients with primary central nervous system lymphoma. Whereas patients without AIDS most commonly seek medical attention because of symptoms suggestive of a discrete space-occupying lesion (isolated neurologic deficits and increased intracranial pressure), patients with AIDS more commonly have more encephalopathic signs and symptoms, characterized by a change of mental status or seizures or both. The global neurologic deficits associated with primary central nervous system lymphoma in AIDS probably result from several factors, including the multifocal nature of this disease, associated infectious processes (HIV and herpes simplex virus, encephalitis, toxoplasmosis, and progressive multifocal leukoencephalopathy), and the greater infiltrative capacity of higher-grade lymphoma cells.

Diagnosis

The diagnosis of primary central nervous system lymphoma may be suggested by several radiographic findings. Primary central nervous system lymphomas are angiographically avascular and either isodense or hyperdense in relation to the normal cerebral cortex on CT scans when compared with most glial tumors and metastases, which are hypodense. Primary central nervous system lymphomas also tend to cause less edema than gliomas or metastases of the same size. More than 90% of primary central nervous system lymphomas are contrast-enhancing, half of which do so homogeneously. Three quarters of all primary central nervous system lymphomas are located adjacent to cortical convexities or ventricular surfaces, probably representing the propensity of this tumor to spread along cerebral spinal fluid pathways. Thus, an avascular mass with less-than-expected cerebral edema, isodense or hyperdense with contrast enhancement on CT scan, located adjacent to a cerebral spinal fluid pathway, is a radiographic image highly suggestive of primary central nervous system lymphoma, especially if the lesions are multifocal (Figure 1).

Although findings of CT scans are remarkably similar in AIDS- and non-AIDS-associated primary central nervous system lymphoma, they have several important differences. The most obvious difference is that multifocal lesions are seen twice as often in patients with AIDS than in the non-AIDS group. Ring enhancement, seen in 50% of all AIDS-associated primary central nervous system lymphomas, is a pattern almost never noted in patients without AIDS (Figure 1). Finally, approximately 10% of all cases of primary central nervous system lymphoma in patients with AIDS are radiographically occult on CT scan. Although MRI scans are probably more sensitive than CT scans for detecting multiple lesions, it remains to be determined whether MRI will be more sensitive in revealing tumors in patients with normal CT scans.

Although the diagnosis of primary central nervous system lymphoma requires histologic confirmation of lymphoma, the type of evaluation may differ in an immunocompetent individual and a patient with AIDS. In the immunocompetent person who is found to have an intracranial mass suggestive of primary central nervous system lymphoma, a tissue diagnosis should be made immediately. If a lumbar puncture can be safely done (no increased intracranial pressure or fourth ventricle obstruction), cerebral spinal fluid should be obtained for both routine and cytologic studies. Although several recent reviews [16, 36] have suggested that the frequency of spinal fluid involvement in primary central nervous system lymphoma is very low (<10%), our examination of the recent literature suggests that it may be as high as 30%. In addition, markers of clonogenicity such as and light chain immunohistochemistry and polymerase chain reaction amplification of B-cell immunoglobulin gene rearrangement may be helpful in distinguishing a clonal proliferation of lymphocytes in the cerebral spinal fluid from reactive pleocytosis [96-98]. Despite these additional tests, most patients will have equivocal cytologic test results and thus require surgical intervention.

Although primary central nervous system lymphoma can be effectively diagnosed by stereotactic biopsy techniques, nearly two thirds of all patients without AIDS during the last 12 years have had open craniotomy and resection rather than biopsy. It has been argued that craniotomy and resection allow a larger pathologic specimen to be obtained, thus ensuring an accurate histologic diagnosis. Recent studies, however, show that stereotactic biopsy provides a high rate of positive tissue diagnosis in primary central nervous system lymphoma, especially when immunohistochemical staining is done (Figure 2). Stereotactic biopsy can be safely done by an experienced neurosurgeon in almost any area of the brain (<2% significant complications), a particularly important point given the deep periventricular location of most primary central nervous system lymphomas. Furthermore, unlike the situation for gliomas, no data suggest that extensive surgery benefits patients with primary central nervous system lymphoma, probably because radiation or chemotherapy or a combination offers such effective cytoreduction. As such, surgery for primary central nervous system lymphoma should be limited to the least invasive procedure that can yield a diagnostic surgical specimen.

After histologic confirmation of primary central nervous system lymphoma, patients should have a full evaluation of the central nervous system (Figure 3, top panel), including a cerebral spinal fluid examination, MRI of the spinal axis, and a slit lamp examination of the eyes because ocular involvement can occasionally occur (approximately 7% of patients at presentation). A common question that arises after the diagnosis of lymphoma in the central nervous system is the extent of staging that is necessary to rule out systemic sites of involvement. From our own experience, as well as from that reported in the literature, a thorough physical examination, routine blood studies, and a chest radiograph are sufficient for screening for systemic lymphoma involvement. If these procedures have normal or negative results, it is unlikely that more extensive tests (gallium scans, body CT scans, bone marrow biopsies) will be useful.

The diagnostic decision algorithm for a patient with AIDS with an intracranial lesion is somewhat more difficult because of the greater possibility that the radiographic abnormality may represent a nonmalignant (progressive multifocal leukoencephalopathy or toxoplasmosis) pathologic process [99, 100]. The decision process is further complicated by the fact that the prognosis of patients with AIDS who have primary central nervous system lymphoma is so dismal that many clinicians feel that a biopsy is not clinically indicated. Although controversy exists about the optimal management strategy in a patient with AIDS who has a CT scan suggestive of toxoplasmosis or lymphoma, a reasonable approach is to treat the patient empirically with antitoxoplasmosis drugs, during which time the patient can be evaluated for bacteremia, fungemia, or other sources of septic emboli. If the patient has not improved clinically or radiographically after 10 to 14 days of empiric therapy, then doing a biopsy should be considered in any patient for whom there is intent to treat should a diagnosis be positive (Figure 3, bottom panel).

Treatment and Prognosis: Non-AIDS-associated Primary Central Nervous System Lymphoma

Radiation Therapy

With the early recognition that surgery alone resulted in only short-term survival (<4 months), radiation therapy became the standard of treatment. Lymphomatous lesions begin to disappear radiographically with doses as low as 2000 cGy, and complete responses are observed in most patients treated with 4000 to 5000 cGy [101, 102]. Despite these positive initial responses to radiation, however, long-term survival is rare with radiation alone [25]. On the basis of retrospective data suggesting a possible radiation dose-survival relation, the Radiation Therapy Oncology Group evaluated a high-dose radiation protocol in 41 patients with primary central nervous system lymphoma using 4000 cGy to the whole brain followed by a 2000-cGy boost to the involved areas. Unfortunately, the 11.6-month median survival of patients treated on this protocol was no better than that of historical controls because 92% of patients, despite the high doses of radiation, developed recurrent disease in sites of previous involvement. This pattern of local failure has been observed in other retrospective reviews of radiation treatment of primary central nervous system lymphoma [36]. Because patients treated on the Radiation Therapy Oncology Group protocol had local failure despite receiving near maximally tolerated cerebral doses of radiation, the limits of radiation therapy were apparent. Furthermore, the addition of spinal axis radiation (except in the rare case of primary spinal intramedullary lymphoma) does not benefit survival because it does not prevent cerebral relapse. Spinal axis radiation will, however, destroy a substantial amount of bone marrow reserve (by radiating portions of the vertebral bodies and pelvis), making subsequent administration of systemic chemotherapy difficult [103].

Chemotherapy: Single Agents

Chemotherapy, shown to be successful in the treatment of systemic lymphomas, has been applied to the management of patients with primary central nervous system lymphoma. The first class of drugs shown to have significant activity against primary central nervous system lymphoma were the glucocorticoids [104, 105]. The lymphotoxic effect of these steroids appears to be much greater in primary central nervous system lymphoma than in systemic lymphoma. Initial response rates of 30% to 70% (some of which are complete responses) have been reported, although these responses are usually short term. The clinician should be aware of the initial effectiveness of steroids in order to expedite a diagnosis before the histologic findings become obscured by the effects of the glucocorticoid agents.

The major theoretical limitation to the effective use of conventional systemic chemotherapy in primary central nervous system lymphoma is insufficient drug delivery because of the blood-brain barrier [106]. Despite this limitation, early anecdotal experience showed that certain chemotherapeutic agents (especially high-dose methotrexate) can effectively treat lymphomatous metastases to the central nervous system [107-110]. Because of these reports and the inadequacy of radiotherapy, investigators have sporadically attempted to use chemotherapy as part of the initial treatment regimen for primary central nervous system lymphoma. Various agents and combinations of agents have been used over the last 12 years. The most commonly used drug was methotrexate, administered by either direct injection into the cerebral spinal fluid (intrathecally or via an Ommaya reservoir) or by intravenous administration. In theory, intrathecal or intraventricular (intracerebral spinal fluid) administration is the preferred method of treating lymphoma cells within the cerebral spinal fluid space and adherent to the meninges (lymphomatous meningitis). Intracerebral spinal fluid therapy does not, however, result in high parenchymal levels of drug in the brain. Because systemically administered methotrexate penetrates an intact blood-brain barrier to an extent proportional to the peak serum level of the drug, high-dose (>1 g/m² body surface area) intravenous methotrexate is preferable to intracerebral spinal fluid administration for treating parenchymal disease. The other two most commonly used drugs (alone or in combination) were high-dose cytosine arabinoside and cyclophosphamide. No single drug regimen was used consistently enough to assess its specific efficacy. Despite this somewhat random use of chemotherapy, a review of the therapeutic results shows a substantial improvement in mean survival (29 compared with 16 months) in those patients treated with chemotherapy compared with those treated with radiation alone.

More important than many of these sporadic case reports, however, are the three large prospective trials evaluating the effectiveness of chemotherapy as part of initial treatment of biopsy-proven primary central nervous system lymphoma in immunocompetent patients (Table 5). In the first trial, Gabbai and coworkers [111] treated 22 patients with three doses of high-dose methotrexate (3.5 g/m²) before radiation. Sixty-seven percent of the patients responded completely (complete resolution of tumor radiographically) to drug therapy alone, whereas most of the remaining patients responded partially. With the median follow-up of 2 years, the median survival has exceeded 27 months. Neuwelt and colleagues [41] reported a prospective trial evaluating the use of cyclophosphamide and methotrexate in conjunction with osmotic blood-brain barrier disruption followed by procarbazine in the initial treatment of patients with primary central nervous system lymphoma. Eighty-one percent of the patients had a complete response to chemotherapy alone, with the remainder having a partial response. The median survival of the 17 initially treated patients was 44.5 months. Of note, 7 patients have remained in remission despite not receiving radiation. Two major questions are raised by this trial. First, is disruption of the blood-brain barrier necessary given that the blood-brain barrier is already substantially disrupted in areas of lymphomatous involvement? The second unresolved question is whether cranial radiation is needed in patients who completely respond to chemotherapy. Neuwelt and colleagues' data suggest that at least some patients may be spared the potential long-term cognitive deficits often associated with cranial radiation. A third prospective chemotherapy trial of primary central nervous system lymphoma was recently reported by DeAngelis and coworkers [18]. Thirty-one patients were initially treated with a combination of intravenous and intraventricular methotrexate, followed by cranial radiation and intravenous cytosine arabinoside. The median survival of this group was 42.5 months. The authors also noted that a cohort of patients with recurrent primary central nervous system lymphoma treated with salvage chemotherapy had significantly prolonged survival compared with control patients with recurrent disease.

The rational design of future regimens will necessitate the use of drugs, at the appropriate dose, that possess both good antilymphoma activity and a relatively good ability to cross the blood-brain barrier. Possible agents include high-dose methotrexate, high-dose cytarabine, procarbazine, cisplatin, nitrosoureas (BCNU, CCNU), thiotepa, and nitrogen mustard. The design of these chemotherapy trials must consider factors such as the widely disrupted blood-brain barrier in areas of lymphoma at the beginning of therapy and subsequent closure of the barrier on tumor regression and barrier repair.

Because patients with intensively treated primary central nervous system lymphoma live longer, investigators will need to address the potential long-term toxic effects of therapy, particularly neurotoxicity. Patients treated with whole-brain radiation often develop long-term cognitive dysfunction. This syndrome of chronic leukoencephalopathy occurs as a result of progressive cerebral demyelination, thought to be the result of radiation-induced microvascular damage. The higher the total dose of radiation, or the larger the dose-fraction size, the greater is the risk for severe cognitive dysfunction. The clinical syndrome of chronic radiation-induced leukoencephalopathy can span the spectrum from mild difficulties with short-term memory to severe dementia. Chemotherapy (particularly methotrexate) has been shown to increase the risk for leukoencephalopathy, at least in children. Retrospective data, however, suggest that the risk for this complication can be decreased by the administration of chemotherapy before radiation and explains the use of chemotherapy before cranial radiation in most of the prospective trials. A question to be addressed in future trials entails the use of preradiation chemotherapy and whether this will allow the use of lower doses of cranial radiation (or eliminate it entirely) in an effort to decrease long-term neurologic sequelae.

In summary, therefore, growing data suggest that the use of chemotherapy as part of the initial treatment of immunocompetent patients with primary central nervous system lymphoma may be changing the natural history of this disease and extending survival. Randomized trials addressing the efficacy of chemotherapy (with or without radiation) over radiation alone are needed. These trials should not only address the effectiveness of therapy but should also evaluate the long-term cognitive effects of therapy by careful prospective neuropsychiatric testing.

Treatment and Prognosis: AIDS-associated Primary Central Nervous System Lymphoma

Unfortunately, the recent advances made in the treatment of non-AIDS-associated primary central nervous system lymphoma have not been translated to advances in AIDS-associated primary central nervous system lymphoma. The overall survival of patients with AIDS-associated primary central nervous system lymphoma is significantly less than non-AIDS-associated primary central nervous system lymphoma (2.6 compared with 18.9 months; Table 4). Even with the use of radiation therapy as the sole treatment modality, the survival differences are remarkable (3 compared with 16.6 months). Several reasons may explain the worse prognosis in patients with AIDS. Patients infected with HIV have an overall poorer response rate to radiation than immunocompetent patients. Whether this fact reflects the possibility that the tumors in the two patient groups are biologically different (more aggressive or radioresistant tumors) or the frequent use of lower doses of radiation in patients with AIDS (56% of patients with AIDS received <3500 cGy compared with only 12% of patients without AIDS) remains unclear.

Although these uncertain differences in tumor biology and approach to treatment may be factors in the poor survival of patients with AIDS-associated primary central nervous system lymphoma, the most important factor is their HIV disease. Levine and coworkers [112] were the first to note that almost all patients with AIDS-associated primary central nervous system lymphoma have CD4+ counts less than 50 cells/mm³. Thus, these patients represent a group with advanced HIV disease. Indeed, a substantial number of patients with AIDS and primary central nervous system lymphoma die of opportunistic infections rather than lymphoma. Although it would be useful to know the specific cause of death in the patients with AIDS reviewed here, those data are not readily available. The best data come from Baumgartner and colleagues [63], who reported the results of autopsies done on 21 patients with AIDS and primary central nervous system lymphoma. All 13 patients who were not treated for their tumors had multifocal lymphoma throughout the central nervous system at the time of autopsy, with 10 patients dying of tumor progression. Of the 8 patients who were irradiated, 2 had residual tumor at autopsy, but only 1 patient died of lymphoma. Thus, 7 of the 8 treated patients died of opportunistic infections. Thus, even successful treatment of the lymphoma may not substantially affect the survival of most of these patients.

Despite the overall poor results of therapy in advanced HIV disease, there are some clinical situations in which treatment should be strongly considered. The first is the HIV-seropositive patient with no antecedent history of opportunistic infections or neoplasms who develops primary central nervous system lymphoma as the AIDS-defining illness. A review of the data suggests that this represents approximately 20% of all HIV-positive patients diagnosed with primary central nervous system lymphoma antemortem. Most series suggest that these patients, treated aggressively, live substantially longer than those with a history of opportunistic infections who subsequently develop primary central nervous system lymphoma. Indeed, the few patients with long-term survival who are described in the literature are those whose primary central nervous system lymphoma was their AIDS-defining illness. These persons are usually in an earlier stage of HIV illness. Generally speaking, this group of patients should be treated with standard radiation with curative intent (5000 to 6000 cGy, whole-brain). The efficacy and toxicity of preradiation chemotherapy in this patient group remains unclear because of the paucity of information on patients with AIDS treated in this manner. We could find reports of fewer than 10 patients with AIDS treated with chemotherapy. The few that have been treated with chemotherapy have received various agents and regimens. Several of these patients have had somewhat prolonged survival (between 6 and 16 months) compared with other patients with AIDS and primary central nervous system lymphoma; however, the numbers are too small to know whether this represents a treatment effect or patient selection bias.

Another group of patients for whom treatment should be considered are those patients who have good performance status and whose major morbidity is related to their primary central nervous system lymphoma. For these patients, radiation therapy offers a relatively nonmorbid, easy treatment that can substantially increase quality of life, even if survival is not appreciably prolonged. Even patients with lymphomatous meningitis may benefit from placement of an Ommaya reservoir and treatment with intraventricular methotrexate or cytosine arabinoside. These treatments can generally be given with little morbidity and on an outpatient basis, forestalling the neurologic sequelae from untreated lymphoma.

It may be most important to distinguish primary central nervous system lymphoma in the patient with AIDS from metastases to the central nervous system from a systemic lymphoma. Because patients with AIDS and systemic lymphoma are often not as severely immunosuppressed (at an earlier stage of HIV disease) as those with primary central nervous system lymphoma, treatment (including chemotherapy) of selected patients with AIDS and systemic lymphoma can result in remission and potential prolongation of survival. Thus, the approach and decision algorithm for these patients may be different than it is for patients with primary central nervous system lymphoma.

Primary central nervous system lymphoma is a disease whose incidence is dramatically increasing in both immunocompetent patients and patients with AIDS. Clinically and biologically, however, primary central nervous system lymphoma appears to differ relative to the host population. For immunocompetent patients, new developments in the effective use of chemotherapy with or without radiation appear to be changing the natural history of the disease and substantially prolonging life. Efforts are under way to develop rational drug and radiation regimens to treat this disease more effectively and to test them in randomized trials. For patients with AIDS-associated primary central nervous system lymphoma, however, the prognosis remains dismal. Although treatment may prolong survival in selected patients and improve neurologic morbidity in others, the overall outcome is poor. Improvement of the prognosis of these patients will first require advances in controlling the progression of HIV disease.