INTRODUCTION — Systemic non-Hodgkin lymphoma (NHL) may involve the nervous system at every level, including peripheral nerve, spinal nerve root, spinal cord, meninges, and brain. Such involvement may include direct invasion or compression of these structures, as well as paraneoplastic effects of NHL. Whereas involvement of the central nervous system (CNS) complicates the course of a substantial minority of patients, peripheral nervous system (PNS) involvement is rare. (See "Overview of paraneoplastic syndromes of the nervous system".)
NHL can involve the CNS either as the sole area of disease (ie, primary CNS lymphoma [PCNSL]) or as secondary spread of systemic disease. Secondary involvement of the CNS by NHL can present in many ways, and rapid control of CNS involvement is necessary to prevent neurologic morbidity and preserve quality of life. As such, clinicians must keep this entity high on their differential diagnosis in patients with NHL presenting with neurologic complaints. The most common manifestations of NHL involvement of the CNS include leptomeningeal disease and parenchymal brain involvement.
The clinical features and diagnosis of secondary CNS lymphoma will be discussed here. Treatment of secondary CNS lymphoma and PCNSL is discussed separately. (See "Secondary central nervous system lymphoma: Treatment and prognosis" and "Primary central nervous system lymphoma: Clinical features, diagnosis, and extent of disease evaluation" and "Primary central nervous system lymphoma: Treatment and prognosis" and "HIV-related lymphomas: Primary central nervous system lymphoma".)
PATHOPHYSIOLOGY — There are limited data regarding the cellular or molecular interactions that facilitate trafficking of lymphoma cells to the CNS. Lymphoma cells are thought to enter the CNS by hematogenous spread, direct extension from adjacent bone marrow infiltration, or centripetal growth along neurovascular bundles. It has also been hypothesized that lymphoma cells could spread from retroperitoneal lymph nodes or bone marrow to the leptomeninges via the intervertebral venous plexus [1], although confirmatory evidence for this mode of spread is lacking.
Most episodes of CNS involvement by non-Hodgkin lymphoma (NHL) occur in the setting of relapsed disease [2]; however, a considerable percentage may have had subclinical CNS involvement at the time of diagnosis. This is suggested by the high rate of CNS involvement within six months of initial therapy in many patients with aggressive NHL [3]. In addition, late CNS recurrences of NHL may represent second primary tumors. Support for this hypothesis comes from molecular studies in which immunoglobulin heavy chain (IgH) gene rearrangement analysis suggested a separate clonal origin for some CNS tumors arising more than three years after the initial diagnosis of NHL [4].
Most data on the mechanism of spread of NHL to the CNS come from studies of varieties of T cell lymphoma with an unusually high rate of extranodal involvement. These malignancies typically express CD56, which is also known as the neural cell adhesion molecule (NCAM) [5]. NCAM is a member of the immunoglobulin superfamily of cell adhesion molecules; it exhibits homophilic (like-to-like) binding, which allows NCAM molecules to facilitate adhesion among neighboring cells with surface NCAM expression. NCAM is naturally expressed on multiple cell types in the brain, spinal cord, muscle, and hematopoietic tissue [5]. As such, expression of NCAM may allow these tumor cells to interact with cells in the CNS that also express NCAM. In a series of 46 patients with T cell lymphoma evaluated during a 12-year period, 24 percent of the tumors were NCAM-positive [6]. Of these, 36 percent had CNS involvement, compared with 3 percent of NCAM-negative tumors. NCAM-positive tumors had a clear predilection for widespread extranodal involvement, with a median survival less than half that of NCAM-negative tumors. (See "Leukocyte-endothelial adhesion in the pathogenesis of inflammation".)
Clues surrounding the migration of NHL to the CNS may also be found by examining the molecular phenotype of normal lymphocytes that migrate to the CNS and how this phenotype differs from that of lymphocytes that do not travel to the CNS [7]. As an example, a study of T lymphocyte trafficking into the subarachnoid space of healthy individuals found that P- and E-selectin expression on endothelial cells in the CNS may be a crucial discrepancy between these lymphocytes and others [8]. Whether these data also apply to neoplastic lymphocytes traveling to the healthy CNS is unclear.
B lymphocytes may also enter the brain even under normal circumstances, although the molecular mechanisms that make this possible have not been fully described [9].
In primary CNS lymphoma (PCNSL), a B cell-attracting chemokine, CXCL13, may play an important role in lymphocyte retention within the CNS [10]. A subset of patients with PCNSL may have occult systemic disease at presentation that eludes detection by routine staging techniques [11]. This patient population may provide insight into the mechanisms by which lymphoma cells home to the CNS. Exome sequencing of PCNSL has revealed frequent mutations in genes known to play a role in CNS development, possibly pointing to mechanisms of CNS tropism [12]. Other research in PCNSL models suggests that faulty germinal center reactions may increase B cell receptor reactivity, particularly for proteins expressed in the CNS, and thereby foster tumor cell survival within the CNS [13].
Another possible homing mechanism is stimulation by posttranslationally modified autoantigens. In one study, two-thirds of patients with PCNSL expressed B cell receptors specific for two autoantigens predominantly expressed in the CNS (neurabin I and atypically hyper-N-glycosylated SAMD14) [14]. By contrast, autoantibodies targeting SAMD14 and neurabin I were not present in patients with secondary CNS involvement by systemic diffuse large B cell lymphoma (DLBCL).
INCIDENCE AND RISK FACTORS — CNS disease is found in a minority of patients at the time of non-Hodgkin lymphoma (NHL) diagnosis, the exact percentage of which differs by histologic subtype. The majority of cases of CNS involvement by NHL occur in the setting of relapsed disease. Nearly half of the patients have concurrent systemic relapse [15].
The frequency of CNS involvement in systemic NHL varies depending at least partially upon the aggressiveness of the NHL subtype. Approximately 2 to 10 percent of patients with aggressive subtypes of systemic NHL (eg, diffuse large B cell lymphoma [DLBCL]) will experience direct involvement of the CNS at some time during their course [3,16-22]. The incidence is much higher in highly aggressive NHL (eg, Burkitt lymphoma, lymphoblastic lymphoma/leukemia) and lower in indolent NHL (eg, follicular lymphoma). In a contemporary multicenter study spanning a decade, the incidence of CNS dissemination by Burkitt lymphoma was 19 percent [23]. Peripheral nervous system (PNS) involvement by lymphoma is rare.
It is unknown whether the risk of CNS relapse has changed as initial treatment of these diseases has evolved. Several retrospective studies have suggested that the incidence may be lower among patients with B cell NHL treated with rituximab-containing therapy [24,25] or etoposide-containing therapy [22,26]. Other studies have found no difference in the incidence of CNS relapse in the pre- versus post-rituximab era [27,28].
In addition to histologic subtype, risk factors for invasion of the neuraxis by NHL include the primary site of disease and the extent of bulky disease (table 1). These issues are discussed in the following sections.
Histologic subtype — The histologic subtype of lymphoma is the major risk factor for nervous system involvement [17,20]. The incidence of CNS relapse without CNS prophylaxis ranges from less than 3 percent in patients with indolent subtypes of lymphoma to 50 percent in some patients with highly aggressive NHL subtypes, such as Burkitt lymphoma or lymphoblastic lymphoma [29]. Although histologic subtype is an indicator of risk for CNS involvement, nearly every lymphoma subtype has been reported to invade the nervous system. (See "Classification of hematopoietic neoplasms".)
●A single-center retrospective study of 2514 consecutive patients with newly diagnosed NHL with a median follow-up of 94 months reported that 106 patients (4.2 percent) developed CNS disease either during (36 patients) or after (70 patients) primary treatment, with a median time to diagnosis of CNS involvement of 10 months [20]. Of the 1163 patients with low-grade histology, 33 (2.8 percent) developed CNS involvement (table 2). Of the 1220 patients with aggressive NHL, 53 (4.3 percent) developed CNS involvement, 35 of which occurred during the first year after NHL diagnosis. This rate has never been confirmed in a prospectively gathered cohort.
●In a retrospective study of 498 patients with NHL, 30 developed secondary CNS disease [17]. Twenty-six of the 30 had aggressive or highly aggressive disease. As examples, the risk of CNS recurrence was 23 percent for lymphoblastic lymphoma and only 1.4 percent for those with indolent lymphoma.
●Mantle cell lymphoma (MCL), which can behave as either an indolent or an aggressive NHL variant, has an unusually high risk of CNS involvement, approaching 4 percent [30,31]. An important factor may be the increased frequency of advanced disease at the time of diagnosis of MCL, when compared with other indolent NHL variants. (See 'Risk scores' below and "Mantle cell lymphoma: Epidemiology, pathobiology, clinical manifestations, diagnosis, and prognosis", section on 'Clinical features'.)
●A single-institution retrospective review of 250 patients with peripheral T cell NHL (PTCL) from the United States reported a CNS relapse rate of 2.4 percent [32]. Another retrospective analysis of 600 PTCL patients revealed a cumulative CNS relapse rate of 2.1 percent at five years [33]. In comparison, a retrospective study of 228 patients from Korea with primary NK/T cell NHL reported a rate of CNS involvement of 8.8 percent [34]. The higher rate in this population likely reflects the predilection of NK/T cell lymphoma for involvement of the hard palate and nose, with venous drainage to the cavernous sinuses.
●Although the indolent T cell lymphoma mycosis fungoides (MF) primarily involves the skin, autopsy and clinical studies have demonstrated nervous system involvement indistinguishable from that caused by other lymphomas [35,36]. (See "Clinical manifestations, pathologic features, and diagnosis of mycosis fungoides", section on 'Clinical features'.)
Primary site — While it is clear that the rate of CNS involvement differs based upon the primary site of disease, there is controversy regarding which primary sites of disease should be considered high risk. Most studies have noted a high rate of CNS involvement with lymphomatous involvement of the testes [22,37,38]. In addition, an increased risk of CNS disease has been associated with lymphomatous involvement of the paranasal sinuses [22,39,40] and retroperitoneal lymph nodes [39,41]. An association between CNS involvement and other sites of primary NHL is more controversial.
●An analysis of CNS relapse among 1386 patients with newly diagnosed aggressive NHL treated on prospective trials of chemotherapy in Germany reported that 2.2 percent of patients developed CNS relapse [22]. When compared with patients without these sites of involvement, the probability of developing CNS relapse by six years was increased in patients with involvement of the testes (22 versus 2 percent), orbit/sinus (33 versus 2 percent), liver, bladder, adrenals, or kidney. Unlike other studies, this analysis did not find an increased risk associated with bone marrow involvement. While these primary sites of disease were associated with increased risk of subsequent CNS involvement, only elevated serum lactate dehydrogenase (LDH) and involvement of >1 extranodal site were independent risk factors for CNS involvement on multivariate analysis.
●Several retrospective studies have noted an increased risk of CNS disease among patients with bone and bone marrow involvement [18,42,43], while others have not [22].
●Several studies have noted an increased risk of CNS relapse among patients with primary testicular lymphoma in both young and older adults [22,37,38,44]. As an example, one review found that 13 of 62 such patients (21 percent) experienced CNS relapse, with 8 of the 13 recurrences isolated to the CNS [37]. Late CNS recurrences were also common, with one patient developing CNS recurrence 13 years after entering remission.
●In a cohort of 87 patients with NHL involving the kidneys identified through a multinational lymphoma study group member survey, the proportion with CNS dissemination was 13 percent (two patients at initial diagnosis, nine at relapse after a median follow-up of 22 months) [45].
●Several studies have noted an increased rate of CNS involvement in patients with primary breast lymphoma [46-50] and primary mediastinal large B cell lymphoma [51]. Whether this risk exceeds the risk in the entire DLBCL population is uncertain [52-54]. An international retrospective analysis of 204 patients with primary DLBCL of the breast, only 8 of whom received CNS prophylaxis, reported CNS relapse in 5 percent after a median follow-up of 5.5 years [55].
●Although primary cutaneous lymphoma is often cited as a risk factor for CNS recurrence, this did not appear to be the case in a large series [56].
●Intravascular lymphoma is an uncommon form of DLBCL with a rate of CNS involvement approaching 40 percent. (See "Intravascular large B cell lymphoma".)
Molecular genetic risk factors — "Double-" or "triple-hit" high-grade lymphomas harboring translocations of MYC and BCL2 or BCL6 carry a high risk of CNS dissemination. In one study, the cumulative risk at three years was 13 percent [57]. An increased risk is also observed in double-expressing lymphoma (MYC and BCL2 expression in the absence of translocations), especially within the ABC subtype (cumulative risk of 10 percent at two years for all double expressers, 15 percent for ABC subtype) [58]. Recognition of these biologic risk factors has led to the use of treatment protocols that include CNS-penetrating drugs. (See "Initial treatment of advanced stage diffuse large B cell lymphoma", section on 'Clinical and laboratory'.)
Other molecular alterations may predict risk of CNS relapse. In one trial cohort of patients with DLBCL, cyclin-dependent kinase inhibitor 2A (CDKN2A) was the most commonly altered gene in patients who developed CNS relapse (mostly homozygous deletions) [59]. On multivariate analysis, CDKN2A gene alterations were associated with higher risk for CNS relapse.
Genetic subtypes defined by next-generation sequencing are also under investigation as predictors of CNS relapse. In one study, a high-risk subtype was characterized by MYD88 and CD79B alterations, similar to those seen in primary CNS lymphoma (PCNSL) [60,61]. A second high-risk subtype contained MYC translocations and BCL2 or tumor protein p53 (TP53) variants.
Risk scores — In patients with DLBCL, the most widely used predictive model for risk of CNS relapse is the CNS International Prognostic Index (CNS-IPI) prognostic score, which was derived and validated from combined data of the British Columbia Cancer Agency and the German High-Grade Lymphoma Study Group [62-64]. The CNS-IPI uses IPI risk factors (age >60, LDH greater than normal, performance status >1, stage 3 or 4 disease, and >1 extranodal site of involvement), as well as the presence of renal or adrenal involvement with lymphoma, to generate a prognostic score with the following risk tiers [62-64]:
●Patients with four to six risk factors had a two-year CNS relapse risk of 10.2 percent
●Patients with two to three risk factors had a two-year CNS relapse risk of 3.4 percent
●Patients with zero to one risk factors had a two-year CNS relapse risk of 0.6 percent
A limitation of the CNS-IPI is that it may not adequately account for anatomic and molecular factors associated with elevated risk of CNS relapse, independent of other clinical risk factors. Aside from renal or adrenal involvement, which is included in the CNS-IPI, DLBCL involvement of the testes and breast also confers increased risk of CNS relapse [37,65-69]. Other potentially high-risk sites include uterus, orbit, retroperitoneum, and bone marrow [20,70,71]. (See 'Primary site' above.)
The combination of CNS-IPI and cell of origin (COO) information (but not BCL2/MYC dual expression status) improved CNS relapse prediction in a retrospective analysis of the large patient cohort enrolled in the GOYA trial [59]. High CNS-IPI score and activated B cell (ABC) type/unknown COO identified a subgroup of patients at high risk of CNS dissemination (15.2 percent at two years versus 0.5 percent for low CNS-IPI score and known non-ABC subtype) [59].
A separate prognostic index has been developed for natural killer cell NHL (CNS-PINK) based on a cohort from the era of non-anthracycline-based chemotherapy with L-asparaginase [72]. CNS-PINK represents a combination of the prognostic index for systemic NK-cell NHL (stratified based on age >60 years, stage III or IV disease, distant lymph node involvement, and nonnasal type disease) and number of extranodal sites. Patients with ≥2 extranodal sites and an intermediate or high PINK score had a cumulative CNS dissemination risk of 23 percent compared with 4 percent in the low-risk group (0 to 1 extranodal sites, low PINK score).
CLINICAL FEATURES — Patients with secondary involvement of the CNS from non-Hodgkin lymphoma (NHL) typically present relatively acutely, with neurologic symptoms emerging over days to weeks. Patients may present with a variety of signs and symptoms based upon the location of the disease within the CNS. In most cases, magnetic resonance imaging (MRI) will show abnormal enhancement within the leptomeninges, brain parenchyma, or spine, and cerebrospinal fluid (CSF) analysis can be diagnostic in cases of leptomeningeal involvement.
Leptomeningeal dissemination — Leptomeningeal spread, or lymphomatous meningitis (LM), is a common nervous system complication of NHL, occurring in as many as 8 percent of patients with NHL [16].
LM typically presents within the first year of diagnosis of NHL. In a large review of NHL patients with CNS disease, a median duration of eight months passed between systemic disease and CNS infiltration [1]. Of the 20 patients with LM, 9 had LM as the initial manifestation of their disease.
Patients with LM commonly present with cranial nerve deficits, radicular pain, vague back or neck pain, mental status changes, focal weakness or sensory loss, or headache. Dysfunction at multiple levels of the neuraxis is common. Seizures are unusual but have been reported [16,18,73], possibly as a result of small tumor deposits causing cortical irritability. Patients may also present with neuroendocrine or behavioral syndromes, such as the syndrome of inappropriate antidiuretic hormone (SIADH) or hypothalamic dysfunction, due to infiltration of specific areas of the brain.
Cranial nerve palsy is among the most common finding, developing in up to 80 percent of patients [16,17]. One or several nerves may be affected, with a predilection for cranial nerves II, III, V, VI, and VII [16,74]. Hydrocephalus with nausea, vomiting, and papilledema may occur [75].
Because lymphoma cells in the subarachnoid space have easy access to every portion of the CNS, LM may present with a multitude of signs and symptoms [76], often lacking classical features of meningeal irritation such as neck rigidity or headache. For this reason, one must have a high index of suspicion for LM when evaluating lymphoma patients. As an example, one case report described hyperventilation as the initial manifestation of LM in a patient with a B cell lymphoma in leukemic phase [77].
Approximately one-quarter of patients with LM involvement will also have intraparenchymal brain lesions [78].
Parenchymal brain metastasis — Parenchymal involvement of the brain by NHL has traditionally been considered distinctly less common than LM, although more contemporary series of patients with secondary CNS lymphoma indicate that the prevalence may be increasing.
In a large series of 592 patients with NHL published in 1980, only 1.4 percent developed cerebral dissemination, representing only 16 percent of the patients who developed CNS involvement of any sort [1]. By contrast, in a multicenter retrospective analysis by the International Primary Central Nervous System Lymphoma Study Group (IPCNSL) that included 92 patients diagnosed from 2000 to 2010 at five different centers, the most common site of CNS involvement was brain parenchyma (43 percent), followed by leptomeningeal (40 percent) and both parenchymal and leptomeningeal (8 percent) [79]. Other studies have reported a similar trend, with approximately 50 to 75 percent of patients having a parenchymal-only relapse [24,25,80-83].
Symptoms and signs depend on the number and location of brain dissemination. Presenting complaints include seizures, focal motor or sensory deficits, cranial nerve deficits, or depressed level of consciousness due to elevated intracranial pressure. (See "Epidemiology, clinical manifestations, and diagnosis of brain metastases", section on 'Clinical manifestations'.)
Spinal cord dissemination — Intramedullary dissemination (IMD) may develop when a tumor in the subarachnoid space grows along nerve roots into the spinal cord or when there is direct hematogenous tumor spread [84]. IMD is sufficiently rare that its prevalence is difficult to estimate, and most patients with IMD present in the setting of widely disseminated disease [85].
In the largest published review of 79 cases of IMD, 6 (7.6 percent) were caused by malignant lymphoma. Lung cancer and breast cancer were the most common primary tumors reported to produce IMD [86]. Although classic neurologic teaching suggests that IMD is more likely than epidural metastases to present with a hemicord (Brown-Séquard) syndrome [84], this has only rarely been reported in the literature [86].
Typical presenting features of IMD include pain, weakness, spasticity, sensory loss, and bowel or bladder dysfunction. These symptoms and signs clearly point to spinal cord disease but are inadequate to distinguish between intramedullary and extramedullary processes [86]. Epidural spinal cord compression secondary to extension of a paraspinal or vertebral NHL mass is discussed elsewhere. (See "Clinical presentation and initial evaluation of non-Hodgkin lymphoma", section on 'Extranodal sites'.)
Neurolymphomatosis — Direct involvement of the peripheral nerves, typically the cranial nerves and spinal nerve roots, is an uncommon complication of NHL. It may occur either as a primary presentation, with or without other sites of systemic or CNS involvement, or as a relapse of prior systemic or primary CNS lymphoma (PCNSL) [87-90].
The majority of patients present with painful radiculopathy involving either cranial or spinal nerves, together with sensory or motor dysfunction involving one or more nerves.
Distinguishing neurolymphomatosis from a toxic or paraneoplastic neuropathy can be challenging in patients with a history of lymphoma and prior chemotherapy, and diagnosis often requires a high index of suspicion, multiple imaging techniques, and nerve biopsy. (See 'Neuroimaging' below and 'Diagnosis' below.)
NEUROIMAGING — Gadolinium-enhanced MRI is the most sensitive imaging technique for detection of CNS involvement from non-Hodgkin lymphoma (NHL). Patients with contraindications to MRI may be evaluated with contrast-enhanced cranial computed tomography.
The following findings may be seen on MRI of the brain or spinal cord:
●Enhancement and enlargement of one or more cranial nerves due to tumor infiltration is extremely suggestive of lymphomatous meningitis (LM) (image 1). High-resolution contrast-enhanced skull base MRI can sometimes show cranial nerve abnormalities that are not seen on routine MRI. Fluid-attenuated inversion recovery (FLAIR) images may show hyperintensity within the subarachnoid space, indicative of high protein content.
●Communicating hydrocephalus, leptomeningeal enhancement, or nerve root thickening are common yet less specific signs of LM (image 2 and image 3 and image 4 and image 5).
●Parenchymal brain deposits from NHL appear as single or multiple contrast-enhancing lesions, often associated with surrounding edema (image 6). In the immunocompetent host, the cellular lesions usually display homogeneously restricted water diffusion on apparent diffusion coefficient maps, distinguishing them from most primary brain tumors and solid-tumor metastases. Typical locations for NHL in the brain include the periventricular white matter and basal ganglia.
●Spinal MRI in LM may show intradural enhancing nodules, especially at the level of the cauda equina; linear enhancement of the leptomeninges is a suggestive finding. Intramedullary spinal metastases appear as contrast-enhancing lesions within the spinal cord itself, often associated with surrounding edema.
●In neurolymphomatosis, MRI of the involved region shows thickening and enhancement of the nerve, plexus, or nerve root.
While positron emission tomography (PET) is generally not useful in the diagnosis of leptomeningeal or parenchymal brain lymphoma, it is indicated in the evaluation of neurolymphomatosis and may be more sensitive than MRI [91]. In such cases, PET may show increased uptake associated with an involved nerve plexus or at multiple levels along the spine within the exiting proximal nerve roots [87,88,92].
LUMBAR PUNCTURE — The classic findings of lymphomatous meningitis (LM) on lumbar puncture (LP) and cerebrospinal fluid (CSF) analysis include a high opening pressure, low glucose concentration, high protein concentration, lymphocytic pleocytosis, and positive cytology and flow cytometry for malignant cells. Although most patients do not have all of these features, an entirely normal CSF examination is uncommon. In particular, the combination of a high protein concentration and lymphocytic pleocytosis is often found without other abnormalities. Radiographic evaluation with MRI of the brain and spine should be performed prior to LP or ventricular tap, although the MRI appearance of LP-induced intracranial hypotension (enhancement and thickening of the dura) is readily distinguishable from radiographic signs of LM.
The site of CSF sampling may affect the ability to detect LM [93,94]. Many patients with suspected LM have Ommaya reservoirs in place that allow for rapid, safe access to ventricular CSF. In one study in which CSF was concurrently sampled from ventricular and lumbar sites, there was a 32 percent discordance rate [94]. For patients with known radiographic disease involving the spinal cord or spinal nerve roots, the diagnostic yield of CSF obtained by LP was higher. Conversely, for patients with known intracranial leptomeningeal disease, the yield of CSF obtained via the Ommaya reservoir was higher. Accordingly, CSF sampling from both sites should be performed when feasible.
The CSF sample should be evaluated for cell counts, protein and glucose levels, cytology, flow cytometry, and immunoglobulin heavy chain (IgH) or T cell receptor gene rearrangement studies. Flow cytometry may detect cases of LM not found by cytology, and cytology may find cases not identified by flow cytometry [95,96].
A variety of molecules present in CSF have been studied as possible diagnostic markers for LM. Examples include beta2-microglobulin [97], D-dimer [98], soluble CD27 [99], and interleukin-10 [100]. None of these have proven adequately reliable or are validated for routine use. Other common but insensitive nonspecific laboratory findings include elevated CSF protein and decreased CSF glucose concentration [74].
CSF cytology — Cytomorphologic examination by light microscopy of a cytospin preparation of CSF is the standard method for determining the presence of lymphoma cells in the CSF. However, false negatives and false positives can occur [101,102]. Because specificity of the technique is substantially greater than sensitivity, false-negative results are more common than false-positive results. A number of shortcomings of cytologic evaluation may preclude rapid diagnosis:
●The cell count in the CSF sample may be inadequate to permit positive identification of malignant cells.
●Infectious causes of atypical lymphocytosis (eg, viral illnesses, Borrelia burgdorferi) may lead to false-positive cytologic diagnoses [103].
●Cellular atypia is common in immunosuppressed patients, increasing diagnostic confusion [104].
●Cytologic analysis requires fresh CSF and adequate preservation of cell structure [102].
Multiple LPs may be required in order to make the diagnosis of LM [102]. In a study of 15 patients with cytology-proven LM, a single LP was adequate for diagnosis in only 60 percent of cases. A second LP detected an additional 33 percent [73]. Other studies have demonstrated that serial LPs increase the sensitivity of CSF cytology for the diagnosis of LM [105,106]; accordingly, most authorities recommend two or three negative LPs when ruling out LM if suspected clinically.
Even when three CSF samples are obtained under optimal circumstances, countless retrospective and anecdotal reports in the literature confirm that cytologic diagnosis may be problematic. Specific guidelines for minimizing false-negative results are discussed separately. (See "Clinical features and diagnosis of leptomeningeal disease from solid tumors", section on 'Cytology'.)
Molecular techniques — Molecular techniques including immunocytochemistry, flow cytometry, and polymerase chain reaction (PCR) can increase the sensitivity of CSF analysis for non-Hodgkin lymphoma (NHL) involvement.
●Immunocytochemistry uses a variety of monoclonal antibodies directed against known cell surface markers and is able to distinguish between lymphomatous and reactive cellular infiltrates. In such cases the immunophenotype of the abnormal CSF cells may be compared with that of the patient's primary tumor [107]. (See "Clinical presentation and initial evaluation of non-Hodgkin lymphoma", section on 'Immunophenotype'.)
●Flow cytometry is a related technique in which clonal populations of B lymphocytes are separated from reactive cells on the basis of size, granularity, and surface antigen expression (figure 1). Flow cytometry can detect populations of neoplastic cells comprising as few as 0.2 percent of CSF lymphocytes [108,109]. There is also a substantial improvement in sensitivity (43 to 50 percent) when flow cytometry is combined with conventional cytology [95,96,108,110-112].
●PCR relies on receptor gene rearrangement during B cell and T cell development to identify clonally related malignant cells (figure 2) [113,114]. As an example, in one study of patients with known CNS lymphoma, five of seven specimens suspicious but nondiagnostic by conventional cytology were positive by PCR [113]. Of 13 specimens that were negative by cytology, 5 were positive by PCR. In another study of patients with CSF involvement by NHL and non-neoplastic lymphoproliferative disorders affecting the CNS, sensitivity of CSF cytopathology, flow cytometry, and IgH gene rearrangement analysis were 27, 10, and 58 percent, respectively. Corresponding specificities were 100, 95, and 85 percent [115].
A single-cell PCR technique designed to detect monoclonality on the basis of IgH gene rearrangement may be more sensitive and specific than standard PCR [116]. However, the primers utilized span only 70 percent of the potential sites of rearrangement; thus, a 30 percent false-negative rate exists [117]. Less experience exists with PCR detection of clonal T cell populations. Comparison of the PCR band size from the CSF with a PCR band from peripheral blood or a lymph node can reduce the false-positive rate.
EVALUATION — Patients with non-Hodgkin lymphoma (NHL) presenting with neurologic signs or symptoms should undergo a complete history and physical examination. The history should pay specific attention to subtle symptoms suggestive of multifocal involvement of the CNS. A careful neurologic examination may reveal findings not suggested by the history.
MRI of the brain and spine should be performed in all patients suspected of having CNS disease. If a routine brain MRI is normal and clinical suspicion is high for leptomeningeal disease, a high-resolution brain MRI may highlight subtle enhancement of the cranial nerves not appreciated on routine sequences. (See 'Neuroimaging' above.)
Patients suspected of having lymphomatous meningitis (LM) as well as those with abnormal parenchymal enhancement suggestive of brain dissemination should undergo lumbar puncture (LP), unless there are contraindications to the procedure (see "Lumbar puncture: Technique, contraindications, and complications in adults", section on 'Contraindications and precautions for high-risk patients'). Cerebrospinal fluid (CSF) analysis should include cytology, flow cytometry, and polymerase chain reaction (PCR) to detect monoclonality on the basis of immunoglobulin heavy chain (IgH) or T cell receptor gene rearrangement. Patients with an Ommaya reservoir in place should have CSF samples collected from both the Ommaya reservoir (ie, ventricular tap) and lumbar space (ie, LP). (See 'Lumbar puncture' above.)
Radiographic evaluation with MRI of the brain and spine should be performed prior to LP or ventricular tap. (See "Lumbar puncture: Technique, contraindications, and complications in adults".)
A complete assessment of systemic disease activity as summarized in the National Comprehensive Cancer Network (NCCN) guidelines for B cell and T cell lymphomas [118,119] is also indicated to facilitate treatment decisions and assist in prognosis.
DIAGNOSIS — A diagnosis of CNS involvement from non-Hodgkin lymphoma (NHL) can be established in several ways:
●A positive cerebrospinal fluid (CSF) cytology establishes the diagnosis of lymphomatous meningitis (LM). Multiple lumbar punctures (LPs) may be required. (See 'CSF cytology' above.)
●Surgical biopsy may be necessary to confirm the diagnosis of parenchymal brain dissemination [120] when CSF analysis is negative and radiographic findings are inconclusive, particularly in patients without a history of prior diffuse large B cell lymphoma (DLBCL) [121]. Imaging features of parenchymal NHL deposits characteristically display homogenous contrast enhancement and relative restriction of water diffusion; often there is infiltration along white matter tracts (image 6). (See 'Neuroimaging' above.)
Because treatment with corticosteroids may delay or interfere with diagnosis and alter imaging characteristics and histopathology of the tumor, it is important that they be avoided before neuroimaging and, if biopsy is indicated, until finalization of biopsy results. If necessary, patients with symptomatic edema or increased intracranial pressure can be initially treated with mannitol.
●For intramedullary spinal metastases, contrast-enhanced MRI in the appropriate clinical context establishes the diagnosis in most cases. CSF cytology will occasionally be confirmatory, particularly in patients with clinical or radiographic suspicion for concomitant leptomeningeal involvement. An open biopsy of the spinal cord is rarely indicated.
●A nerve biopsy is typically required for diagnosis of isolated neurolymphomatosis. CSF cytology is usually negative. In the appropriate clinical context (ie, a patient with a known history of NHL presenting with pain and neurologic deficits referable to multiple nerves), a characteristic pattern of nerve root thickening and enhancement or positron emission tomography (PET)-avid nerve roots may be sufficient.
DIFFERENTIAL DIAGNOSIS — A variety of conditions can mimic the clinical presentation or MRI appearance of leptomeningeal dissemination from non-Hodgkin lymphoma (NHL), including infections, inflammatory processes, or late effects of prior therapy (table 3). The differential diagnosis of leptomeningeal dissemination is discussed in more detail elsewhere. (See "Clinical features and diagnosis of leptomeningeal disease from solid tumors", section on 'Differential diagnosis'.)
For patients presenting with a brain mass, the differential diagnosis includes other neoplasms as well as non-neoplastic entities such as vascular lesions or infection (table 4). (See "Overview of the clinical features and diagnosis of brain tumors in adults", section on 'Neuroimaging features'.)
SUMMARY AND RECOMMENDATIONS
●Systemic non-Hodgkin lymphoma (NHL) may involve the nervous system at every level, including peripheral nerve, spinal nerve root, spinal cord, meninges, and brain. Such involvement may include direct invasion or compression of these structures, as well as noninvasive paraneoplastic effects of NHL. (See 'Pathophysiology' above.)
●Central nervous system (CNS) disease is found in a minority of patients at the time of NHL diagnosis, the exact percentage of which differs by histologic subtype. The majority of cases of CNS involvement by NHL occur in the setting of relapsed disease within the first year after completion of therapy. (See 'Incidence and risk factors' above.)
●Leptomeningeal dissemination, or lymphomatous meningitis (LM), is the most common nervous system complication of NHL. Patients with LM commonly present with radicular pain, vague back or neck pain, mental status changes, cranial nerve deficits, focal weakness or sensory loss, or headache. Dysfunction at multiple levels of the neuraxis is common. (See 'Leptomeningeal dissemination' above.)
●Parenchymal involvement of the brain by NHL has traditionally been considered distinctly less common than LM, although the prevalence may be increasing. Presenting complaints include seizures, focal motor or sensory deficits, cranial nerve deficits, or depressed level of consciousness due to elevated intracranial pressure. (See 'Parenchymal brain metastasis' above.)
●Intramedullary spinal dissemination (IMD) is a rare manifestation of secondary CNS lymphoma that may develop when a tumor in the subarachnoid space grows along nerve roots into the spinal cord or when there is direct hematogenous tumor spread. (See 'Spinal cord dissemination' above.)
●Gadolinium-enhanced MRI is the most sensitive imaging technique for detection of CNS involvement from NHL. All patients with suspected CNS disease should undergo brain and spine MRI prior to lumbar puncture (LP). (See 'Neuroimaging' above and 'Evaluation' above.)
●The diagnosis of lymphomatous meningitis (LM) is made based upon the pathologic evaluation of a sample of cerebrospinal fluid (CSF) interpreted within the clinical context. The CSF sample should be evaluated for cell counts, protein and glucose levels, cytology, flow cytometry, and immunoglobulin heavy chain (IgH) or T cell receptor gene rearrangement studies. Flow cytometry may detect cases of LM not found by cytology, and cytology may find cases not identified by flow cytometry. (See 'Lumbar puncture' above.)
●NHL lesions are distinguished from other neoplasms and infectious, inflammatory, or metabolic causes by homogenous contrast enhancement, relative restriction of water diffusion, infiltrative appearance, and characteristic location. Surgical biopsy may be necessary to confirm the diagnosis. (See 'Diagnosis' above.)
●The differential diagnosis includes a variety of conditions that can mimic the clinical presentation or MRI patterns seen in patients with suspected LM from NHL, such as infections, inflammatory processes, late effects of prior therapy, and imaging artifacts. (See 'Differential diagnosis' above.)
ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges Andrew Norden, MD, and Ephraim Hochberg, MD, who contributed to an earlier version of this topic review.
Do you want to add Medilib to your home screen?