Management of Primary Central Nervous System Lymphomas

Published on 13/03/2015 by admin

Last modified 13/03/2015

Print this page

This article have been viewed 2730 times

Chapter 12 Management of Primary Central Nervous System Lymphomas

The phrase “primary central nervous system lymphoma” (PCNSL) is used to designate an extranodal lymphoma restricted to the nervous system, which account for about 3% of all brain tumors. Most are large B-cell lymphomas but a few cases of T-cell lymphomas have been reported. A common location is the brain parenchyma surrounding the ventricular system, but any craniospinal structure, in addition to the eye, can be involved. Although not as common, isolated spinal cord, meningeal, or ocular PCNSL can also occur. On the other hand, brain lesions are not infrequently accompanied by leptomeningeal and ocular dissemination. This chapter does not cover nervous system involvement as the first manifestation of systemic lymphoma, which can masquerade as PCNSL.

Among brain tumors, PCNSL has gained notoriety because, although still rare, it has recently increased in incidence and, unlike other brain tumors, has a high response rate to chemotherapy and radiation therapy. Before 1980, PCNSL would occur in a few individuals who were immune suppressed, usually after kidney transplant. The advent of the acquired immunodeficiency syndrome (AIDS) epidemic brought a steep increase in the frequency of this tumor. Nevertheless, the increased incidence was also seen in individuals without AIDS or other known immunosuppressive states, except for older age. Pathogenesis, diagnostic approach, treatment, and prognosis differ according to the patient’s immune state; thus when there is a suspicion of PCNSL, establishing an individual’s immunocompetency is of fundamental importance in deciding the most appropriate management.

At some point in his or her career, the neurosurgeon will be required to decide about surgery for a lesion suspected of being PCNSL by imaging studies. Unfortunately, the appearance is not specific, making it necessary to have this entity in mind as part of the differential diagnosis of any mass lesion. Because this tumor is highly responsive to nonsurgical forms of therapy, the role of surgery has to be tempered accordingly. In some cases, it will entail the deferral of surgical resection of a mass until the pathologic result of a diagnostic biopsy is available for review. In others, it involves refraining from the use of steroids until after the biopsy is performed to ensure the best diagnostic yield from the specimen. Moreover, placement of a reservoir with intraventricular catheter for chemotherapy administration may be required as part of the treatment. Therefore, it is important for the neurosurgeon to be aware of the clinical and diagnostic characteristics suggestive of PCNSL while actively participating in the subsequent therapeutic antineoplastic phase.

This chapter describes concepts pertaining to PCNSL of relevance for the neurosurgical practice taking into account, where appropriate, differences according to the patient’s immune state. Initial consideration will be given to the pathogenesis, etiology, and epidemiology of this type of lymphoma. Clinical presentation and diagnostic approach when the suspicion of PCNSL arises will be reviewed. Finally, therapeutic interventions, complications, and prognostic factors will be described in detail, insofar as they are important to understanding the surgical role in the overall interdisciplinary treatment approach of PCNSL.

Pathogenesis and Molecular Pathology

PCNSLs are almost exclusively of B-cell origin with only 2% of T-cell origin. The most common histologic subtype is the CD20-positive, diffuse, large-cell, B-cell, non-Hodgkin’s lymphoma (NHL), with a smattering of other more indolent B-cell lymphomas reported. The disease is more common in the immunocompromised than in the immunocompetent, but the pathogenesis of these disorders is uncertain regardless of the immunocompetency of the patient.^1,²

Normal B cells arise from the hematopoietic stem cell and initially undergo antigen-independent differentiation, with immunoglobulin rearrangement in the bone marrow prior to emerging from the marrow as mature but naive B cells. These B cells move to secondary lymphoid organs where, upon encountering antigen, they undergo somatic hypermutation of the immunoglobulin variable region in the germinal center microenvironment. The presence of T cells, antigen presenting cells and the appropriate cytokine milieu are generally considered a requirement for somatic hypermutation with subsequent affinity maturation.³ B cells displaying the highest affinity for antigen are rescued from apoptosis and become either a memory cell or the terminally differentiated plasma cell.

Malignant B cells can be viewed as B cells arrested at a certain stage of differentiation. The developmental state of the cell will be reflected in its morphologic attributes, the degree of immunoglobulin rearrangement, the expression of surface molecules including CD10, BCL-6, MUM-1, and CD138 (which serve as markers of the B cell’s transition through the germinal center), and the presence of intraclonal heterogeneity. In the majority of PCNSLs, the malignant cells are BCL-6 positive (60%–100%), MUM-1 positive (90%–100%), and have undergone immunoglobulin rearrangement with somatic hypermutation. These findings suggest that the malignant cell has seen antigen, passed through the germinal center but has not yet become a plasma B cell, that is, the cell of origin in most PCNSLs is a B cell on the verge of exiting the germinal center (Fig. 12-1).⁴^–⁶ Because there are no germinal centers in the brain, the B cell has, in all likelihood, migrated from a node to the CNS, probably in response to an antigen.⁷ It is unclear, however, whether the malignant transformation of B cell occurs prior to or following migration into the CNS.

FIGURE 12-1 A model for the histogenesis of PCNSL based on the developmental stage of B-cell arrest as indicated by antigen expression.

(Adapted from Camilleri-Broet S, Criniere E, Broet P, et al. A uniform activated B-cell-like immunophenotype might explain the poor prognosis of primary central nervous system lymphomas: analysis of 83 cases. Blood 2006;107:190-196.)

A rat model of the disease, developed by Knopf and colleagues,⁸ suggests that PCNSL arises in response to an antigenic stimulus, an infection perhaps, where the antigen has moved into a draining lymph node and serves to recruit naive B cells. Presumably, antigen retained in the CNS as well as the expression of specific chemokines prompts trafficking of B cells back into the CNS. Although this hypothetical scenario is compatible with the pathologic stage of development and differentiation of the malignant cell, several questions remain unanswered, including identification of the site of malignant transformation as mentioned above, the complete lack of involved lymph nodes, and the identification of the intracerebral antigen driving the process.

Certainly, there is abundant data that Epstein-Barr virus (EBV) is involved in the pathogenesis of PCNSL in the HIV-positive patient. There are, however, equally compelling data that EBV positivity is a rarity in PCNSL occurring in the immunocompetent individual.^3,^9,¹⁰ Thus, it is quite possible that the underlying pathogenesis differs, depending on the immunocompetency of the patient.

Epidemiology

PCNSL was considered a rare tumor occurring in a few immunosuppressed organ transplant recipients until the early 1980s when, coinciding with the AIDS epidemic, there was a marked increase in its frequency. The increase in incidence was seen in all age groups but was more evident in men than in women.¹¹ At its peak, prior to the highly active antiretroviral therapy (HAART) era, the relative-risk in HIV-infected individuals was 1000- to 3600-fold higher than in immunocompetent individuals; this has declined dramatically, essentially by an order of magnitude, since the introduction of HAART.^12,¹³ However, a definite and persistent rise in incidence in the immunocompetent population has been observed.^14,¹⁵ Patients without immune suppression are usually older, and the male-to-female ratio is 1.2 to 1.7:1.¹¹ Most studies corroborate that this change in the incidence of PCNSL is independent of trends in the incidence of brain tumors and in NHL.¹⁶

Clinical Manifestations

The clinical effects of PCNSL are indistinguishable from those associated with other brain tumors. In addition to the routine medical history, special care should be taken to elicit information about the possibility of immune suppression, especially HIV infection. PCNSL usually occurs several years after the diagnosis of HIV infection has been made.¹⁷ Because PCNSL only occurs in about 3% of all AIDS patients,¹⁸ infections like toxoplasmosis are a more likely diagnosis in this setting. It should be noted, however, that as a cause of intracranial mass lesions in an individual with AIDS, PCNSL is second only to toxoplasmosis.¹²

Approximately 8% of immunocompetent patients will have a history of successful treatment of a non–nervous system malignancy prior to the diagnosis of PCNSL.¹⁹ In these cases, the diagnosis is even more challenging and might be delayed because of the concern of secondary nervous system involvement from the previous malignancy. When the previous malignancy is an extraneural NHL, absence of systemic disease on diagnostic workup and comparison of gene rearrangement studies on the biopsy specimens of both lymphomas can demonstrate that they are separate entities. Whether these patients have an increased predisposition to multiple neoplastic processes, or the PCNSL is the result of the antineoplastic treatment for the first tumor, is unknown.¹⁹

The relative frequency of clinical manifestations of PCNSL does not differ greatly between immunosuppressed and immunocompetent individuals. Nevertheless, there are some differences that might be of clinical relevance when considering the diagnosis. In immunocompetent patients, the median age at presentation is in the sixth decade of life, whereas the median age for AIDS patients is in the fourth decade of life. Patients with AIDS more often have multiple lesions than do immunocompetent individuals, making the clinical topographic diagnosis difficult, and the latency between the onset of symptoms and diagnosis seems to be shorter.¹⁸

The clinical course is usually subacute, with a few months elapsing between the onset of symptoms and the diagnosis of a mass lesion by imaging studies. There are several reports of spontaneous transient remission of symptoms associated with PCNSL.²⁰ The most common symptoms associated with PCNSL are focal neurologic deficit, increased intracranial pressure, alteration of mental function, or a combination of these manifestations. At the time of initial presentation, approximately one third will have symptoms of increased intracranial pressure, about 50% will have behavioral changes, and approximately 10% will have seizures.¹⁸ Between 30% and 42% of patients will experience a combination of focal and nonfocal symptoms at the time of diagnosis. The neurologic examination will yield a variety of signs that can be localizing or nonlocalizing (e.g., increased intracranial pressure or alteration in the mental function). Hemiparesis and ataxia are the most common focal neurologic signs, but aphasia, acalculia, visual field defects, and cranial nerve palsies are also common.¹⁸ Cranial-spinal nerve palsies and hydrocephalus might be secondary to lymphomatous meningeal infiltration, which is present in up to 42% of all patients with PCNSL.¹⁸ Visual symptoms might precede or follow the diagnosis of PCNSL and will depend on the ocular structure affected by the tumor. However, 50% of the patients with PCNSL and ocular involvement detected by slit-lamp examination are asymptomatic.²¹ Clinically apparent ocular involvement at presentation is found in about 8% to 10% of PCNSL patients,¹⁸ but vitreous involvement of the eye occurring prior to or during the course of CNS lymphoma has been noted in up to 25% of patients.²² In about half of the cases with ocular involvement, visual symptoms can be the first manifestation of PCNSL, preceding neurologic symptoms by several months. Decreased visual acuity or floaters may prompt the patient to seek medical attention, and any nonspecific uveitis refractory to topical or systemic steroids should bring ocular PCNSL to mind.²¹

Rare clinical syndromes that sometimes are associated with PCNSL include those where the tumor location is restricted to the spinal cord, the leptomeninges, and the hypothalamus. These are especially challenging cases because, in addition to other neoplastic diseases, benign inflammatory entities can have a similar clinical and radiologic presentation. Isolated spinal cord involvement occurs in 1% to 2% of all PCNSL ¹⁸ and can be associated with syringomyelia.²³ The level and extent of myelopathic involvement will guide the clinical presentation. Secondary involvement of the spinal cord in patients with cerebral lesions is not a rare occurrence.²⁴ PCNSL may also present as hypothalamic dysfunction causing diabetes insipidus,¹⁸ as pituitary apoplexy with bitemporal hemianopsia or as isolated third nerve palsy.²⁵

A variant of PCNSL, clinically presenting with progressive cognitive decline and gait disorder and associated with diffuse white matter abnormality without enhancement on MRI was initially described in 1999.²⁶ The term “lymphomatosis cerebri” has been used to describe this uncommon condition, which can be difficult to diagnose because of the nonspecific clinical and imaging findings;²⁶^–³⁰ it can be erroneously diagnosed as vascular leukoencephalopathy, multiple sclerosis (MS), or gliomatosis cerebri.

Diagnosis

A definitive diagnosis of PCNSL cannot be made on clinical or imaging grounds, and histologic confirmation is essential. CT scan or MRI will initially establish the presence of a mass lesion with characteristics that, although suggestive of PCNSL, are not specific for this entity (Fig. 12-2). CSF analysis helps in the differential diagnosis and in some cases makes the diagnosis by the demonstration of malignant B lymphocytes.¹⁸ HIV testing is required in all patients suspected of having PCNSL. In spite of the information obtained from these studies, tissue diagnosis is required in most circumstances.

FIGURE 12-2 T1-enhanced (A) and FLAIR (B) MRI showing the infiltrative character of PCNSL. Arrow shows small satellite area of enhancement that on the FLAIR seems to be connected to the larger lesion.

Imaging Studies

There are some characteristics on imaging studies that, although not pathognomonic, would strongly suggest that the mass lesion identified might be of lymphomatous origin. In the immunocompetent host a higher level of suspicion is required. Head CT scan shows a hyperdense or isodense mass, solitary in 86% of the cases,³¹ that usually exhibits homogenous enhancement after the administration of iodinated contrast.³² Lesions are usually supratentorial and localized in the deep periventricular areas.³¹ Because of their infiltrative nature, the lesions might have indistinct borders and result in minimal surrounding edema or compressive effect (see Fig. 12-2). Occasionally it can appear as a ring-enhancing lesion with a hypodense, necrotic core, indistinguishable from a high-grade glioma. Lesions can be multiple, suggestive of metastatic disease or infection, but the scarce perilesional edema should raise the suspicion of PCNSL. In about 10% of the cases the lesions are localized in the posterior fossa.

On MRI, most of the lesions are hypointense or isointense on T1-weighted images, and only about 40% are hyperintense on T2-weighted images.³² Although most lesions enhance after the administration of gadolinium exceptions include cases when the study follows the administration of steroids or when there is a diffuse infiltrating lymphoma (lymphomatosis cerebri) (Fig. 12-3). The appearance of PCNSL is of such heterogeneity that it should be considered in the differential diagnosis of any mass lesions detected on imaging studies (Fig. 12-4).

FIGURE 12-3 FLAIR MRI prior to (A) and after (B) treatment of lymphomatosis cerebri.

FIGURE 12-4 T1-enhanced MRI of entities mimicking PCNSL on imaging studies: (A) multiple sclerosis, (B) toxoplasmosis in a patient with a history of renal transplant, (C) stroke, and (D) arteriovenous malformation.

PCNSL in the immunocompromised patient may have a more variable appearance on imaging studies.³³ Infection is more likely to be responsible for a mass or enhancing lesion on imaging studies than PCNSL in this population, but no specific pattern has been established that can be used to distinguish between CNS lymphoma, toxoplasmosis, or other CNS diseases that occur in patients with AIDS.¹⁸ Unlike immunocompetent patients, AIDS-associated PCNSL has been reported to present with multiple lesions in 71% to 80% of cases, to show ring-like enhancement in 50% of cases, and to lack enhancement in about 10% to 27% of the lesions.³³ Spontaneous hemorrhage, a nonenhancing lesion, or diffuse white matter changes do not exclude lymphoma in an immunocompromised patient.

Advanced MRI techniques may be helpful to pre-operatively suggest PCNSL. Magnetic resonance spectroscopy (MRS) shows massively elevated lipid resonances in PCNSL; although this may also be present in glioblastoma, the finding of elevated lipid resonances in combination with a markedly elevated choline/creatine ratio, may improve the preoperative differentiation of PCNSL and glioma.³⁴ Dynamic susceptibility contrast-enhanced perfusion MRI reveals that the relative cerebral blood volume ratio is lower for PCNSL, correlating with the lower microvessel density by immunohistochemistry, than high grade gliomas.³⁵ Using diffusion tensor imaging, with the enhancing lesion as the region of interest and comparing to the contralateral normal-appearing white matter, the fractional anisotropy and the apparent diffusion coefficient (ADC) ratios can be measured. Both parameters are significantly lower in lymphomas than glioblastoma and may assist in differentiating between these two entities.³⁶ The ADC ratios, however, do not allow a reliable distinction between toxoplasmosis and lymphoma.³⁷

Nuclear medicine has also been suggested as a possible method to discriminate between PCNSL and other types of malignant as well as nonmalignant pathologies without resorting to histologic diagnosis. In a study comparing N-isopropyl-p-(¹²³I)-iodoamphetamine (¹²³I-IMP) single photon emission computerized tomography (SPECT) in patients with malignant glioma, PCNSL and meningioma, the ¹²³I-IMP retention uptake in the 6-hour to 24-hour SPECT images were significantly higher in PCNSL than in those of both malignant gliomas and meningiomas.³⁸ In patients with AIDS, the thallium²⁰¹-SPECT delayed retention index may be useful to discriminate PCNSL from infectious lesions with high sensitivity and specificity.³⁹ The diagnostic utility of these techniques has yet to be determined in larger series, and histologic confirmation is still considered the gold standard for diagnosis of PCNSL.

Tumor infiltration of the nervous system can be more diffuse than is appreciated on imaging studies. Correlation of autopsy and MRI findings in 10 patients who died with PCNSL showed that all had tumor infiltration in CNS regions that were normal radiographically, including T2 sequences.⁴⁰ Therefore, the infiltrative microscopic tumor burden of PCNSL renders futile any attempt to resect these lesions. The surgical role is restricted to a tissue biopsy for histologic diagnosis.

Tissue Diagnosis and Staging

While the diagnosis of PCNSL is usually suggested by the appearance of a focal lesion on CT or MRI, confirmation of the diagnosis of PCNSL in the immunocompetent requires tissue to definitively differentiate PCNSL from metastatic disease, glioma, sarcoidosis, and inflammatory lesions (Table 12-1). Because the CSF is involved at diagnosis 20% of the time, a brain biopsy can be avoided if malignant cells can be obtained from the CSF.⁴¹ Similarly, malignant (though often asymptomatic) uveitis is apparent in 10% to 20% of patients with PCNSL at presentation and may serve as the source of cells on which the diagnosis may be based.^42,⁴³ Cytologic examination of cells, as the sole parameter by which the diagnosis is made, is not optimal because it cannot determine monoclonality, a condition that is necessary though not sufficient for the diagnosis of lymphoma. In the case of B-cell lymphoma, monoclonality can be determined by flow cytometric detection of light-chain restriction if there are sufficient cells available. In T-cell disease, and when there are inadequate numbers of malignant B cells for flow cytometry, the diagnosis can be established by polymerase chain reaction (PCR) studies on the basis of T-cell receptor or immunoglobulin rearrangement, respectively. This technique is susceptible to false positives if too few cells are present and to false negatives if the DNA is highly degraded.⁴⁴ It is important to underscore the fact that most lymphomas, including PCNSL, are quite sensitive to steroids. Cell death and tumor regression may occur as early as 24 hours after initiation of therapy. If tissue or a specimen for cytology is obtained following initiation of steroids, the result may be nondiagnostic because of cell death and tissue necrosis. Unless the patient is showing evidence of rapid neurologic deterioration or impending herniation, tissue should be obtained prior to starting any steroid therapy.^45,⁴⁶ Although a retrospective study exists challenging this concept,⁴⁷ it seems prudent to avoid the unnecessary use of corticosteroids prior to diagnostic biopsy for PCNSL.

Table 12-1 Differential Diagnosis of PCNSL

Disease	Diagnostic Studies
Multiple sclerosis	Past medical history, CSF
High-grade glioma	SPECT-MRS
Infection	HIV infection, CSF
Sarcoidosis	ACE level, calcium
Meningioma	MRS
Vascular	Cerebral angiogram, MRI DWIs

ACE, angiotensin-converting enzyme; CSF, cerebrospinal fluid; DWI, diffusion-weighted images; HIV, human immunodeficiency virus; MRS, magnetic resonance spectroscopy; SPECT, single photon emission computerized tomography.

Once the diagnosis is made, patients generally undergo staging to determine the extent of involvement. Baseline evaluations include a physical and a neurologic exam, as well as cognitive function assessment. Because systemic involvement tends to occur more commonly at relapse, and because evidence of systemic disease is found in less than 5% at diagnosis, the necessity for a full lymphoma staging has been called into question.⁴⁸ There is general agreement that, besides a complete history and physical, a CBC, standard chemistries with an LDH, HIV status, chest x-ray, CSF examination, and slit-lamp examination are absolutely required. A full staging would also include CT of the chest, abdomen, and pelvis, as well as bilateral bone marrow biopsies. Finally, testicular ultrasound should be considered in all elderly men. Full staging is inevitably required for any patient enrolled in a clinical trial and is recommended for all patients with PCNSL in the published guidelines for standardization of baseline evaluations.⁴⁹

The role of ¹⁸F-fluorodeoxyglucose (FDG) PET to rule out systemic disease in the initial evaluation of patients presenting with PCNSL is uncertain. Two retrospective studies using PET in the initial staging of immunocompetent patients have revealed abnormalities (e.g., other malignancies, systemic sites of lymphoma as well as concurrent sites within the nervous system) undetected by other diagnostic studies in 19% and 28% of patients respectively.^50,⁵¹ It remains to be determined if FDG PET is needed in all patients with the presumptive diagnosis of PCNSL (Fig. 12-5).

FIGURE 12-5 Coronal T1-enhanced MRI (A) and 18-fluoro-2-deoxyglucose (FDG) PET image (B) of patient with PCNSL at the time of diagnosis.

The approach to a focal brain lesion in the HIV population is slightly different. The detection of EBV DNA by PCR in the CSF was initially found to be a sensitive (80% to 84%) and highly specific (100%) diagnostic marker of AIDS-related PCNSL, thus potentially obviating the need for a biopsy; these results, however, could not be confirmed in a later study.⁵²^–⁵⁴ A brain biopsy in the AIDS patient carries a significant morbidity (8.4%) and mortality (2.9%),⁵⁵ but is recommended when the CSF is negative for lymphoma, the focal lesion is atypical for toxoplasmosis, there is progression on a brief trial of antitoxoplasmosis therapy, toxoplasma serologies are negative, there is a rapid neurologic decline, or there are discordant CSF EBV and thallium²⁰¹-SPECT results.^12,¹⁷ It is noteworthy that a pre-HAART retrospective study of presumed or confirmed PCNSL in HIV-positive patients revealed there was no difference in the overall survival (1.2 months) between the two groups, presumptive and biopsy-confirmed, leading the authors to conclude there may be little benefit in subjecting the patient to the diagnostic procedure given the dismal outcome.⁵⁶ The use of HAART and its impact on the CD4 count has allowed the use of more aggressive chemotherapeutic regimens without undue toxicity, and resulted in improved life expectancy for these patients. These changes will undoubtedly influence the algorithm, and they suggest that earlier diagnostic procedures may be in order for this patient population.