ABSTRACT: Primary CNS lymphoma is rising in incidence in both the AIDS and non-AIDS populations. It is a non-Hodgkin's lymphoma that usually presents as a brain tumor, but the leptomeninges, eyes, and spinal cord also are frequently affected. Systemic lymphoma is not present, and comprehensive systemic staging is unnecessary, but appropriate neurologic staging is imperative. Standard therapy has been whole brain radiotherapy, giving a median survival of 12 to 18 months in non-AIDS patients, but only 2 to 5 months in AIDS patients. In non-AIDS patients, the addition of chemotherapy to radiotherapy has improved the prognosis, with median survivals of 30 to 45 months. Current protocols focus on the development of combination chemotherapy programs and reducing the dose of cranial radiotherapy to minimize late neurologic sequelae. The addition of chemotherapy to brain irradiation prolongs survival in some patients with AIDS-related disease, but median survival is not significantly improved. [ONCOLOGY 9(1):63-78, 1995]
Primary central nervous system (CNS) lymphoma is a non-Hodgkin's lymphoma that arises within and is restricted to the nervous system. It usually presents as a brain tumor, but may also involve the leptomeninges, eyes, spinal cord, or any combination of these sites . Formerly, the disease was rare, representing less than 1% of all intracranial tumors, but since 1974 there has been at least a threefold rise in incidence in apparently immunocompetent patients. It has also been recognized for decades that primary CNS lymphoma occurs with increased frequency in patients with congenital (eg, ataxia-telangiectasia) or acquired (eg, renal transplant recipients) immunodeficiencies. It is now most common in patients with the acquired immunodeficiency syndrome (AIDS), as many as 6% of whom develop primary CNS lymphoma during their illness. It is so characteristic of the immunosuppression associated with AIDS that it is classified as an AIDS-defining illness in HIV-infected individuals.
Historically, treatment of primary CNS lymphoma was surgery followed by cranial irradiation; the median survival was 12 to 18 months, and the 5-year survival rate, 3% to 4% in immunocompetent patients; immunosuppressed individuals fared worse. Survival figures with this treatment approach were consistent in numerous retrospective series evaluating small numbers of patients (20 to 30) collected over several decades. These studies also demonstrated the aggressive biologic behavior of primary CNS lymphoma, with untreated patients surviving only a few months. Furthermore, the collected series suggested that surgical resection had no significant therapeutic role in the treatment of this disease, because survival was not worsened if only biopsy for tissue diagnosis was performed. Because the disease was so infrequent, pursuit of more effective therapy seemed neither practical nor necessary.
In the early 1980s, physicians began to see a rising number of immunocompetent patients with primary CNS lymphoma, an increase later confirmed by epidemiologic data. During this time, the full force of the AIDS epidemic led to increased cases of AIDS-related primary CNS lymphoma, and recent studies demonstrate that the incidence continues to rise, as HIV-infected individuals survive longer [2,3]. These factors clearly highlighted the inadequacy of then-current therapy, and clinical trials were begun in an effort to improve treatment.
To design an appropriate therapeutic approach, a number of issues concerning the biologic behavior of primary CNS lymphoma needed to be defined and incorporated into the treatment plan. Using the clinical staging criteria applied to systemic non-Hodgkin's lymphoma, primary CNS lymphoma is a stage IE lymphoma; that is, it involves a single extranodal site, the brain. Most patients with systemic stage Ie non-Hodgkin's lymphoma can be treated effectively with involved field irradiation. Focal radiotherapy achieves a complete response rate of almost 100% and can yield 10-year survival or cure rates of 70% in patients with stage I disease . Using involved field radiotherapy (whole brain radiotherapy) for the treatment of primary CNS lymphoma does not produce comparable control of disease. Two possible reasons are that (1) disease exists outside the radiotherapy field and that (2) pathologically, primary CNS lymphoma has more aggressive biology than identical systemic lymphomas.
The issue of dissemination of disease is more easily addressed than the question of a fundamental biologic difference between primary CNS lymphoma and systemic lymphomas, but systemic and nervous system dissemination must be considered independently.
Systemic Dissemination Not a Factor
It has been hypothesized that an occult systemic lymphoma is the origin of primary CMS lymphoma, because the nervous system has no lymph nodes or lymphatics, and thus the source of malignant lymphocytes in the CNS must be from outside. If an occult systemic lymphoma accounts for disease in the nervous system, then primary CNS lymph- oma would simply be a stage IV lymphoma with an unusual and rare pattern of metastasis to CNS only. However, the evidence argues strongly against this hypothesis.
First, when systemic lymphoma spreads to the nervous system, it spreads to meninges and affects the brain in only 1% of patients . In those few patients with brain metastases, the lesions appear late in the course of their illness and are associated with recurrent or uncontrolled systemic disease, often specifically with bone marrow infiltration or retroperitoneal adenopathy.
Second, patients with primary CNS lymphoma have no evidence of systemic lymphoma, either at diagnosis or at autopsy. At Memorial Sloan-Kettering Cancer Center, more than 100 non-AIDS patients with primary CNS lymphoma had a complete systemic evaluation at diagnosis, including body computed tomography (CT) scans, bone marrow aspiration, and biopsy, and not one had evidence of systemic lymphoma. There are two case reports of patients who presented with apparent primary CNS lymphoma and had systemic lymphoma on evaluation, but one patient had two histologically different lymphomas, and the disease in the second patient likely represented systemic metastases from untreated primary CNS lymphoma [6,7].
Autopsy studies have found systemic lymphoma in only 7% to 8% of primary CNS lymphoma patients, and most of these patients have a single site of microscopic disease believed to be a metastasis from their uncontrolled CNS tumor. Rare patients with primary CNS lymphoma develop clinically obvious systemic lymphoma, which may even develop in the context of undetectable cerebral tumor. In some of these patients, the systemic lymphoma may be a second primary, as primary CNS lymphoma is seen occasionally in patients with a variety of prior systemic neoplasms ; however, the development of systemic lymphoma is very unusual and does not represent the typical clinical behavior of primary CNS lymphoma.
These data demonstrate that systemic lymphoma is not a factor in the natural history of primary CNS lymphoma, and that treatment need not include therapy designed to eradicate occult systemic disease.
Nervous System Dissemination
Although systemic dissemination is not a factor when considering treatment of primary CNS lymphoma, dissemination within the nervous system must be considered (Table 1). Primary CNS lymphoma is confined to a single extranodal site, the central nervous system, but is almost always disseminated within it. This dissemination can take many forms and involve different compartments of the CNS. The brain is the most important site of disease and is involved in more than 95% of patients . By magnetic resonance (MR) scan, parenchymal brain lymphoma is multifocal in about half of patients at diagnosis. However, autopsy and some radiographic studies have demonstrated that the contrast-enhancing lesions seen on scan can significantly underestimate the extent of tumor present within the parenchyma . At autopsy, all patients have microscopic infiltration involving regions of brain and occasionally spinal cord that were radiographically normal. Consequently, the tumor burden is always greater than initially appreciated, and if enhancement on CT or MR scan measures blood-brain barrier integrity, all patients have some disease behind a relatively intact blood-brain barrier.
The Brain—The brain is the most important area of involvement by primary CNS lymphoma, but dissemination within other CNS compartments, such as the leptomeninges, eyes, and spinal cord, may also be important, and appropriate neurologic staging, including lumbar puncture and ophthalmologic examination, is imperative (Table 2). Primary CNS lymphoma tends to form mass lesions in periventricular locations. It is a highly infiltrative neoplasm and will usually traverse the ependyma to involve the ventricular surface. Lesions more peripherally located grow into the overlying leptomeninges. Leptomeningeal infiltration is so characteristic of primary CNS lymphoma that Shibata10 has postulated that malignant lymphocytes from the subarachnoid space are the source of primary CNS lymphoma and grow into the brain to form parenchymal mass lesions.
The Leptomeninges—In many patients, leptomeningeal involvement remains localized to areas adjacent to the parenchymal lesions, accounting for the absence of clinical symptoms or signs specifically suggestive of leptomeningeal tumor, such as cranial neuropathies or radiculopathies (!-Figure 1). In spite of the relatively localized leptomeningeal involvement seen pathologically, malignant cells are identified in the cerebrospinal fluid (CSF) of one third of patients at diagnosis; in an additional third, lymphocytes with a suspicious cytology are present . Thus, there is a high incidence of at least microscopic CSF dissemination of primary CNS lymphoma at diagnosis.
In occasional patients, widespread leptomeningeal lymphoma with clinical evidence of subarachnoid tumor may be the presenting problem, but these patients are a distinct minority. Nevertheless, most patients have tumor within the leptomeninges, at the gross or microscopic level, and thus specific therapy must be directed against the CSF.
The Eyes—In addition to the leptomeninges, the eye is another important site of primary CNS lymphoma [12-14]. Embryologically and functionally, the eye is an extension of the nervous system. Occasionally, ocular involvement (as opposed to orbital lymphoma, which always represents systemic disease) occurs by direct extension of brain disease through the optic nerve or meninges, but more commonly it arises as an independent site of this multifocal process . Patients may present with blurred vision or floaters, but clinically this may be hard to distinguish from more common benign ocular conditions, particularly in an older population. In the AIDS population, these symptoms may be confused with ocular infections, particularly cytomegalovirus (CMV) retinitis. In addition, about half of patients with ocular lymphoma have no visual symptoms.
About 12% to 18% of patients in whom primary CNS lymphoma began in the brain have ocular involvement at diagnosis.11 Conversely, lymphoma can begin in the globe, and 50% to 80% of patients with ocular lymphoma develop cerebral involvement during the course of their disease. Frequently, the latency from onset of ocular lymphoma to the appearance of CNS disease is many years. Ocular lymphoma is detected by slit-lamp ophthalmologic examination. Vitrectomy may demonstrate malignant lymphocytes, but such a biopsy may yield a false negative if the patient has been taking corticosteroids .
While the relationship between ocular and cerebral lymphoma has been known for years, ophthalmologic examination has only recently become part of the staging evaluation of patients with primary CNS lymphoma, and, therefore, clinically silent ocular infiltration was not detected previously. This has important therapeutic implications, since ocular involvement necessitates specific therapy, which is not administered in the course of standard whole brain radiotherapy. Neglect of ocular lymphoma not only results in progressive ocular disease with visual loss but also can lead to CNS relapse and an apparent failure of standard treatment.
The Spinal Cord—Parenchymal spinal cord involvement is rare and usually clinically evident; however, we have encountered a few patients with spinal cord infiltration missed at diagnosis and not covered by the initial therapy. Complete imaging of the spinal cord is not necessary in every patient with primary CNS lymphoma, but patients with neck or back pain, and symptoms or signs suggestive of myelopathy should have an enhanced MR scan of the entire spine. Very rarely, primary CNS lymphoma originates within the spinal cord, but localization of the disease is usually straightforward, since patients present with progressive paraplegia .
In virtually all patients, primary CNS lymphoma is disseminated throughout the nervous system at diagnosis. While almost all patients present with brain tumor(s), many will have involvement of the eyes, leptomeninges, spinal cord, or some combination of these sites. These regions must be incorporated into a treatment regimen or they will serve as a potential reservoir of untreated disease that will lead to subsequent relapse. Nevertheless, failure to cover sites of dissemination accounts for treatment failure in only about 10% of patients.
In about 90%, primary CNS lymphoma relapses occur in the brain, often at sites remote from the original tumor location. Some brain recurrences may be attributed to progression from ocular or meningeal sites, but uncontrolled brain lymphoma is the most likely source for the majority of relapses. Relapse within the irradiated field is common after focal radiotherapy for primary CNS lymphoma, unlike stage IE systemic lymphoma, indicating that primary CNS lymphoma is a more biologically aggressive disease than comparably staged systemic non-Hodgkin's lymphoma . No molecular features responsible for this biologic aggression have been identified.