Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Diffuse Large B-Cell Lymphoma of the Central Nervous System
Diffuse large B-cell lymphoma (DLBCL) of the central nervous system (CNS) accounts for an overwhelming majority (95%) of cases of primary CNS lymphoma (PCNSL). According to the 2016 report of the World Health Organization (WHO), it is defined as DLBCL arising within the brain, spinal cord, leptomeninges, or eye. The WHO definition excludes lymphomas of the dura, intravascular large B-cell lymphomas (IVLBL), secondary CNS lymphomas (SCNSLs), and all immunodeficiency associated lymphomas from this category ( Table 84.1 ). For the purpose of discussion, in this text, the terms PCNSL and DLBCL of the CNS are used interchangeably. The diagnosis and management of primary vitreoretinal lymphoma (PVRL) is also discussed, which is a subset of PCNSL. In the context of human immunodeficiency virus (HIV) infection, PCNSL is one of four acquired immunodeficiency syndrome (AIDS)-defining illnesses with distinct clinicopathologic characteristics and unique therapeutic challenges. Secondary CNS lymphoma (SCNSL) may occur as a recurrence of previously diagnosed non-Hodgkin lymphoma (NHL) or may occur simultaneously as a manifestation of systemic disease. They can metastasize to the brain parenchyma, but more often, they spread to leptomeninges, causing symptoms across the entire neuraxis, including epidural cord compression in approximately 3% to 5% of patients. Risk recognition and the prevention of CNS disease are important aspects of the overall management strategy in high-grade systemic DLBCLs.
Table 84.1
2016 World Health Organization Classification of Lymphomas of the Nervous System
|
AIDS , Acquired immunodeficiency syndrome; CNS , central nervous system; DLBCL , diffuse large B-cell lymphoma; EBV , Epstein-Barr virus; NK , natural killer; NOS , not otherwise specified.
The primary origin of PCNSL is unclear, given the virtual absence of B cells in the normal brain tissue. The disease is categorized mainly as activated B cell (ABC) or non-germinal center B (non-GCB). However, the cell of origin (COO) distinction has uncertain therapeutic and prognostic merits in PCNSL. More importantly, central to CNS lymphogenesis is the constitutive activation of B-cell receptor (BCR) signaling pathway, with its downstream target, nuclear factor-κB (NFκB), affected by frequent recurrent mutations, mainly MYD 88 and CD79B. from a diagnosis standpoint, the inherent challenges in timely diagnosis of PCNSL have paved the way for numerous minimally invasive diagnostic biomarkers, albeit with limited success, in turn supporting biopsy of the affected area as the current diagnostic gold standard.
There is a paucity of high-quality comparative data to guide practitioners with level 1 evidence-based treatment recommendations for PCNSL or SCNSL. Nonetheless, unlike most other brain tumors, PCNSL often responds favorably to systemic therapy, specifically high-dose methotrexate (HD-MTX) based (poly) chemotherapy and/or whole-brain radiation therapy (WBRT). The latter modality, however, is plagued by unsustainability of responses and debilitating neurotoxicity, particularly in the older adults. Following successful induction therapy, consolidation is administered to deepen and/or maintain the response. Overall, the treatment program of induction and consolidation attains long-term disease-free survival (DFS) in approximately 40% to 60% of the patients underscoring the fact that about half or more of the patients would relapse within the neuraxis.
Most recently, the concept of maintenance therapy has emerged, especially for patients who are unsuitable for consolidation. Novel approaches and agents are also being investigated and will be discussed throughout the chapter. The prognosis of patients who relapse, or are refractory, is dismal and treatment remains less optimally defined than for newly diagnosed PCNSL. Whenever possible, patients with PCNSL and SCNSL should be guided immediately towards a well-designed, appropriately powered clinical trial.
Epidemiology
PCNSL accounts for less than 2% of all NHLs, 4% to 6% of all extranodal lymphomas, and 3% to 6% of all brain tumors, with an age-adjusted incidence rate of 4 cases per million persons per year. Approximately 1500 new cases are diagnosed each year in the United States, a rate that is steadily increasing as the population ages. The median age of diagnosis is 56 years in an immunocompetent host, with about an equal male-to-female ratio. PVRL is the most common form of intraocular lymphoma (IOL). Approximately, 15% to 25% of patients with parenchymal PCNSL subsequently develop “secondary” vitreoretinal lymphoma (VRL), underscoring the importance of a bilateral eye exam at diagnosis and periodically thereafter. Conversely, about 65% to 90% of patients with PVRL consequently have or eventually develop parenchymal PCNSL, underscoring the importance of brain surveillance imaging. Primary choroidal lymphomas do not have association with CNS disease. Secondary choroidal lymphomas represent an extranodal manifestation of systemic lymphoma and are usually confined to the choroid. The most common systemic lymphoma subtype, involving the choroid, is DLBCL, followed by multiple myeloma, extramedullary plasmacytoma, Waldenström macroglobulinemia, and chronic lymphocytic leukemia (see Chapter 90, Chapter 92, Chapter 67 respectively).
Besides advancing age, congenital and acquired immunodeficient conditions are other well-established risk factors for PCNSL ( Table 84.2 ). PCNSL, in immunocompromised individuals, occurs at an earlier age with a higher predilection for males than females. It is less clear whether it follows a more aggressive course in this patient population. In an HIV-infected individual, PCNSL is an AIDS-defining disease, associated with the Epstein-Barr virus (EBV) infection (see Chapter 78, Chapter 87 ). The EBV association in an immunocompetent host with PCNSL is less frequent (<15%). The risk factors of PCNSL in HIV-infected individuals are a low CD4 count (<100,000 cells/mm 3 ) and a high viral load. Due to effective antiretroviral therapy, the incidence of HIV-associated CNS lymphoma has been declining.
Table 84.2
Immunodeficiency and Primary Central Nervous System Lymphoma Incidence
| Condition | Life-time risk of PCNSL |
|---|---|
|
|
|
|
|
|
AIDs , Acquired immunodeficiency syndrome; HIV , human immunodeficiency virus; PCNSL , primary central nervous system lymphoma.
PCNSL may also present as a second malignancy. For example, it should be considered in the differential diagnosis of patients with “recurrent disease of solid tumor” that is exclusively confined to the neuraxis or in situations where the primary systemic tumor is inactive and the patient develops a new brain lesion.
Pathogenesis
Despite significant progress in the elucidation of the genomic landscape of PCNSL, a specific genetic defect has not been reported. Due to its rarity and location, access to the copious tissue for diagnosis (and research) is a major limitation. Furthermore, prediagnostic and nonjudicious use of steroids in a majority of immunocompetent patients with brain mass further restricts the quantity, as well as quality, of the examined neural tissue.
Given the virtual absence of B cells in the healthy brain parenchyma, the origin of the malignant transformation of PCNSL remains a mystery. Similar to systemic DLBCL, PCNSL is also likely to originate from late GC-derived B cells as suggested by the ongoing somatic hypermutations (SHMs) in immunoglobulin (Ig) genes. Moreover, frequent (50% to 80%) usage of the V4–34 gene segment suggests that specific antigens trigger the development of PCNSL. Presently, no cell adhesion molecule or chemokine predicting B cell homing selectively to the brain has been identified in the evolution of PCNSL.
Interestingly, concomitant small-size monoclonal B-cell populations have been identified in the peripheral blood and/or marrow in a subset of patients (~15%) with PCNSL. In a study by Fukumura et al., individuals with tumors positive for MYD88 mutations also harbored the same mutations, albeit at a low frequency in peripheral blood mononuclear cells, suggesting that MYD88 mutation-positive precancerous cells originate outside of the CNS. Hypothetically, these cells might develop additional genetic hits that confer adaptation to the CNS environment. However, such subclinical extra-CNS clonal positivity by itself does not imply presence (or absence) of systemic lymphoma.
Additionally, a plethora of highly sophisticated experimental, albeit small studies using comparative genomic hybridization (CGH) or single-nucleotide polymorphism (SNP) array karyotyping and whole-exome sequencing (WES) techniques have highlighted myriad and heterogeneous cellular and microenvironment abnormalities, forming the contexture of PCNSL. These abnormalities involve genes influencing chromatin structure and modification, cell-cycle regulation, adhesion molecules and immune recognition with a remarkably high prevalence of oncogenic aberrations in BCR and Toll-like receptor (TLR), triggering brisk NF-κB signaling in the majority of patients with PCNSL ( Fig. 84.1A,B ) ( Chapter 90, Chapter 92, Chapter 67, Chapter 90, Chapter 90 ).
Genomic Landscape and Aberrant Pathways
The pathogenesis of this multicompartment disease is fragmented and unclear, with a few exceptions. The CGH has identified many nonspecific genetic abnormalities. Chromosome 6q deletions, in particular the 6q21–23, may be the most frequent alteration. They occur in about 40% to 60% of PCNSL. Candidate tumor suppressors linked to chromosome 6q include: PRDM1, a tumor suppressor and regulator of terminal B-cell differentiation; PTPRK, a protein tyrosine phosphatase that participates in cell adhesion signaling events; and A20 (TNFAIP3), a negative regulator of NFκB signaling. Along with 6p21 deletion, PCNSLs frequently exhibit 9p24.1/PD-L1/PD-L2 copy number alterations and translocations, a likely genetic explanation underlying immune evasion within this immunopriviledged site. Interestingly, Four et al. reported high levels of PD-L1 expression with nominal PD-L1/PD-L2 translocations in 32 immunocompetent PCNSL patients. Recurrent chromosomal losses have also been detected in the 9p21 site, which encodes loci involved in cell cycle regulation, including CDKN2A, a putative target for cyclin-dependent kinase inhibitors. Recurrent gains have been detected on chromosome 12, specifically in the 12q region harboring STAT6, MDM2, CDK4 and GLI1 genes. Abnormalities in the long arms of chromosomes 1, 7, and 18 have also been reported.
The BCR, TLRs, and NF-κB pathways are frequently activated due to genetic alterations, affecting in particular (see Fig. 84.1A and B ): MYD88, CD79A/B, INPP50 (also called SHIP), CBL, BLNK, CARD11, MALT1, and BCL2, which foster proliferation and prevent apoptosis. Other genes that have been implicated in the pathogenesis include: IRF2BP2, TBL1XR1, ETV6, IRF4, IRF2BP2, EBF1, CCND3; and CDK20, CREBBP, MLL2, ARID1A/B, and SMARCA4. A recent study by Bödör et al. utilized NanoString Lymphoma Subtyping Test (LST)-assay gene expression in formalin-fixed paraffin-embedded (FFPE) tissue. They elegantly showed that the most frequently mutated genes in the PCNSL cohort ( n = 64), included MYD88 (66%), PIM1 (41%), KMT2D (31%), and PRDM1 (30%). The mutation frequencies in the remaining genes were as follows: MYC (19%), IRF4 (19%), CD79B (17%), TP53 (11%), CCND3 (9%), CARD11 (8%), PAX5 (3%), CSMD2 (3%), and CSMD3 (3%). No mutation was found in PTPRD gene. In the SCNSL cohort ( n = 12) the PRDM1 (50%), followed by MYD88 (42%) and PIM1 (25%) were the most frequently mutated target genes. The mutation frequencies in the remaining genes proved to be lower compared to PCNSL: KMT2D (17%), CD79B (8%), IRF4 (8%), CCND3 (8%), C-MYC (8%), TP53 (8%), and PAX5 (8%). No mutation was identified in CARD11, CSMD2, CSMD3, and PTPRD genes.
In another recent study by Nayyar et al., phylogenetic analysis of paired primary and relapsed specimens from patients with PCNSL identified MYD88 mutation and loss of CDKN2A as early clonal events. In a study by Balint et al., the NGS analyses revealed that over 80% of potentially protein-changing mutations in lymphomagenesis were located in eight genes: CTNNB1, PIK3CA, PTEN, ATM, KRAS, PTPN11, TP53, and JAK3. The TP53 was the only gene harboring mutation seen in all 19 patients. Inactivation of p14 ARF and p16 INK4a genes has also been implicated in pathogenesis. Somewhat at odds with these findings are results from the study by Chapuy et al. which reported the rarity of TP53 mutations. However, the p53 pathway was frequently perturbed upstream by way of a mono- or bi- allelic loss of CDKN2A. As a consequence, it is suggested that certain patients with such a mutation may benefit from MDM2/4 inhibitors that may augment wild-type p53 activity and CDK-blockade.
MicroRNA Expression
The discovery of microRNAs (miRNAs) has opened a new field for unraveling and therapeutically targeting diseases (see Chapter 4 ). These small noncoding RNAs regulate diverse biological processes through posttranscriptional gene expression modulation. A study by Fischer et al. demonstrated a distinct pattern of miRNA expression in PCNSL ( n = 11) when compared to systemic DLBCL ( n = 10), using paraffin-embedded biopsy specimens in immunocompetent patients. In all, 18 miRNAs were differentially expressed between PCNSL and systemic DLBCL. The upregulated miRNAs in PCNSL specimens were associated with the MYC pathway (miR-17-5p, miR-20a, miR-9), with the blocking of terminal B-cell differentiation (miR-9, miR-30b/c), or with upregulation by inflammatory cytokines (miR-155). Putative tumor-suppressor miRNAs (miR-199a, miR-214, miR-193b, miR-145) were downregulated in PCNSL. There was no overlap of miRNAs dysregulated in PCNSL with those differentially expressed between IHC, defined as GCB and non-GCB types, or in those apart from miR-9, with miRNAs known to be overexpressed in human brain. Notably, the results of another study by Robertus et al. are somewhat contradictory, in which they reported that miR-155 showed the lowest expression level compared with other miRNAs involved in PCNSL. This could be possibly explained by a different sample preparation, varying RNA quality, and low sample size in both studies.
Adhesion and Chemokine Molecules
The exact mechanism behind CNS tropism and dissemination within the CNS is not clear but may be facilitated by the increased expression of extracellular matrix molecules such as LFA-1/ICAM-1, CD44, Fas (CD95), a transmembrane receptor protein, and B cell attracting chemokines such as CXCL-12 (SDF-1) and CXCL-13 that bind to CXCR4 and CXCR5, respectively, on lymphoma cells (see Fig. 84.1B ) (see Chapter 90, Chapter 92, Chapter 67, Chapter 90 ).
Immune Evasion and Its Implications
Loss of both human leukocyte antigen (HLA) class I and class II expression in B cell lymphomas is a mechanism of escape from a cytotoxic T lymphocyte (CTL) immune response in lymphomas of immune-privileged sites. HLA-A, HLA-B, HLA-C, and HLA-DR are variably expressed, with approximately 50% of CNS DLBCLs showing loss of major histocompatibility complex (MHC) class I and/or II expression (see Chapter 24 ). Interestingly, Alama et al. retrospectively analyzed PCNSL transcriptomes in 54 samples (sequencing, n = 20; microarrays, n = 34) from PCNSL. Integrated correlation analysis and signaling pathway topology enabled them to infer intercellular interactions. Immunohistopathology and digital imaging were used to validate the bioinformatic results. Transcriptomics allowed them to classify the tumors into three immune subtypes: immune-rich, poor, and intermediate. The immune-rich subtype was associated with better survival and characterized by hyper-activation of STAT3 signaling and other proinflammatory signaling (e.g., interferon [IFN]-γ and tumor necrosis factor [TNF]-α), resembling the hot subtype described in primary testicular lymphoma and solid tumor. A group of developmental signaling pathways including WNT/β-catenin, HIPPO (Salvador-Warts-Hippo), and NOTCH were hyper-activated in the immune-poor subtype. HLA down-modulation was associated with a low or intermediate immune infiltration and the absence of T-cell activation. Moreover, HLA class I down regulation was also correlated with worse survival, with implications for immune-intermediate PCNSL that frequently feature reduced HLA expression. A ligand–receptor intercellular network revealed a high expression of two immune checkpoints (i.e., CTLA-4/CD86 and TIM-3/LAGLS9). Mucin-domain containing protein-3 (TIM-3) and galectin-9 proteins were clearly upregulated in PCNSL (see Chapter 58 ). Based on their observations, they suggest testing DNA methyltransferase (DNMT) inhibitors in PTPN6 methylated PCNSL (observed in 48.5% of their study samples) to restore STAT3 expression in immune-infiltrated PCNSL, ibrutinib in combination with immune checkpoint inhibitor in immune rich/MYD88-mutated PCNSL patients (ClinicalTrials.gov Identifier: NCT03770416), and histone deacetylase (HDAC) inhibitors that crosses blood brain barrier (BBB) to re-establish the HLA class II expression in immune-poor PCNSL subtype. Miyasato et al. reported that the expression of PD-1 ligands on tumor-associated macrophages (TAM) was dependent on the activation of STAT3. In addition, the expression of indoleamine 2,3-dioxygenase (IDO1) by TAMs was involved in the immune evasion by lymphoma cells in patients with PCNSL, suggesting that combination therapy using dual inhibitors of IDO1 and immune-checkpoint may have a therapeutic value. The increased incidence of IDO expression in a normal “healthy” brain with aging may help explain the increased risk of PCNSL in older patients. Marcelis et al. showed high expression of PD-L1 in 28% of PCNSL without amplification of the 9p24.1 locus. The TME was predominantly composed of CD8+ T cells and macrophages. The presence of the former cells was associated with an improved overall survival (OS). On the contrary, there was a correlation between high expression of TIM-3 in T cells in the TME and inferior outcome. The investigators reported marked intertumoral and intratumoral heterogeneity in the TME.
Cell of Origin and Its Implications
The precise implication of COO in PCNSL remains unclear. Data using the Hans classifier show both GCB and non-GCB subtypes, but unlike nodal DLBCL (see Chapter 85 ), the COO IHC classification does not affect progression-free survival (PFS) or OS. Limitations of these studies include small sample sizes, retrospective nature, and incomplete treatment information that could impact survival analysis. On gene expression profiling (GEP), with the limitation of sample-introduced bias, the majority of PCNSL are non-GCB subtype. Interestingly, Bödör et al. demonstrated a higher proportion of GCB (13% vs. 5%) and a significantly lower proportion of ABC (80% vs. 95%) PCNSL cases when compared with parallel Hans characterization. These findings are in contradiction with previous reports. Such discrepancies most likely can be explained by the heterogeneity in the type and depth of the sequencing approaches (whole genome/exome sequencing vs. targeted resequencing of selected genes), the type of the analyzed material (fresh frozen vs. FFPE), and the difference between the bioinformatics pipelines and variant calling methods. The same study also showed that mutations of TP53 (15% vs. 6%) and PAX5 (15% vs. 2%) were more frequent in GC cases, compared to ABC cases, respectively. While mutations of CD79B, CARD11, CSMD2, and CSMD3 were exclusively detected in ABC cases, MYD88, PRDM1, and KMT2D genes showed similar mutational frequencies across the two subtypes. This finding may be of potential therapeutic relevance, as the BTK inhibitor ibrutinib has demonstrated clinical efficacy in 37% (14 of 38 patients) of nodal ABC-DLBCLs, especially in cases with concurrent MYD88 and CD79B mutations. In a phase I study by Grommes et al., clinical responses were seen in 10 of 13 patients with relapsed PCNSL who had MYD88 mutation. Importantly, this study showed that CARD11 mutations were associated with ibrutinib resistance.
Moreover, Montesinos-Rongen et al. analyzed tissues from 21 patients with PCNSL using a microarray-based expression profile. A comparison of the transcriptional profile of PCNSL with various normal and neoplastic B-cell subsets demonstrated PCNSL (i) to display gene expression patterns most closely related to late GCB cells, (ii) to display a GEP similar to systemic DLBCLs and (iii) to be in part assigned to the ABC or the GCB subtype of DLBCL. Interestingly, their study supported the notion that PCNSL exhibits characteristics associated with both ABC (IGM expression with absence of IG class switch, high frequency of somatic mutations of IG genes, recurrent 18q21 gains, activation of the NF-κB pathway, and high expression of BCL2 mRNA and protein) and GCB type (expression of BCL6, ongoing somatic hypermutation) DLBCL. These data, along with the data from Fukumura et al. and Inoue et al. support that PCNSL represents a distinct COO entity irrespective of the conventional Hans COO classification. Additionally, a more recent, large Canadian population-based study of 155 patients, suggests that PCNSL has a unique molecular profile distinct from systemic DLBCL.
Epstein-Barr Virus–Associated (Tissue-Positive) Primary Central Nervous System Lymphoma and Its Implications
EBV is most frequently observed in HIV-associated PCNSL and early posttransplant PCNSL (see Chapter 87 ). A recent study by Gandhi et al. showed that the genetic landscape typically observed in EBV − HIV − PCNSL is distinct from EBV-associated PCNSL. They used targeted sequencing and digital multiplex gene expression to compare the genetic landscape and TME of 91 DLBCL histology cases. Forty-seven patients were EBV tissue-negative: 45 EBV − HIV − PCNSL and 2 EBV − HIV + PCNSL; and 44 were EBV tissue-positive: 23 EBV + HIV + PCNSL and 21 EBV + HIV − PCNSL. As noted previously, EBV − HIV − PCNSL had frequent MYD88, CD79B, and PIM1 mutations, and enrichment for the ABC subtype. In contrast, these mutations were absent in all EBV-associated PCNSL, and ABC subtype frequency was low. Furthermore, copy number loss in HLA class I/II and antigen-presenting/processing genes was rarely observed, indicating retained antigen presentation. To counter this, EBV + HIV − PCNSL had a tolerogenic TME with elevated macrophage and immune-checkpoint gene expression, whereas AIDS-related PCNSL had low CD4 gene counts. Interestingly, the proportion of PD-L1 and PD-L2 gene amplification was similar between EBV − HIV − PCNSL and EBV + HIV − PCNSL, suggesting that the increase in expression of these ligands in EBV + HIV − PCNSL may be due to macrophage expression. from a therapeutic standpoint, checkpoint blockade is potentially contraindicated in post-transplant lymphoproliferative disease (PTLD) due to risk of graft rejection and graft-versus-host disease. By contrast, checkpoint blockade might be a reasonable option in AIDS-related PCNSL. Adoptive transfer of EBV-specific third-party virus specific T cells chosen on the basis of the best HLA match and in vitro effector function see ( Chapter 90, Chapter 93 ) has previously been shown to induce high response rates in EBV + PTLD and can cross the blood-brain barrier (see Chapter 90, Chapter 93 ). Based on the cumulative biological data, a phase I Australasian Leukemia/Lymphoma Group clinical trial incorporating EBV-specific third-party virus specific T cells (ACTRN12618001541291) has commenced.
Clinical Presentation
The clinical presentation of CNS lymphoma depends on the size, pattern, and location of the disease within the neuraxis.
Brain Parenchyma
The lesions vary in size from microscopic implants to large bulky masses. PCNSL often presents in a subacute manner with chronic lethargy, nonspecific decline in cognition, and psychomotor function (32% to 43%), often delaying the diagnosis. Focal neurologic deficits (56% to 70%) can be observed immediately prior to the diagnosis. Presentation of headache, confusion, diplopia, nausea, and vomiting are usually late signs and symptoms of elevated intracranial pressure (ICP). Seizures are uncommon (14%), mainly due to non-cortical location of the disease. The classic B symptoms are rarely observed in PCNSL.
Primary Vitreoretinal Lymphoma
A proportion of PCNSL patients present with PVRL. Overall, visual symptoms are rare at presentation, even though the ocular involvement is relatively high (15% to 20%). Insidious and nonspecific onset of symptoms can often delay the diagnosis of PVRL by 6 to 24 months. PVRL symptoms can imitate a wide range of other ocular entities and have been described as a masquerade syndrome. Patients may be asymptomatic or complain of blurred vision (40% to 50%), reduced visual acuity (25% to 30%), and vitreous floaters (20% to 25%), often misdiagnosed as inflammatory eye conditions such as uveitis, vitritis, chorioretinitis, or normal degenerative changes. The clinical signs are bilateral in approximately 60% to 90% of individuals. A history of an older adult with noninfectious uveitis who failed to respond to anti-inflammatory therapy should raise a suspicion of VRL. Other important differential diagnoses include fungal or bacterial endophthalmitis, sarcoidosis, ocular syphilis, and tuberculosis. The classic B symptoms are rarely observed in PCNSL.
Leptomeninges
Simultaneous involvement of leptomeninges is observed in a small subset (18%) of PCNSL patients, but exclusive meningeal presentation is extremely unusual. The presence of leptomeningeal involvement at diagnosis should increase suspicion of SCNSL. Symptoms of lymphomatous spinal cord lesions are parallel to those observed with other intramedullary tumors and depend on tumor location. In general, thoracic spine is the most common area of involvement by PCNSL.
Neuroimaging
Up to 75% of all PCNSLs contact a cerebrospinal fluid (CSF) surface: either the ventricular ependymal or pia lesions are often deep-seated with a predilection for the periventricular white matter, especially the corpus callosum. The basal ganglia and thalami are the next most common locations. Occasionally, PCNSLs diffusely infiltrate large areas of the hemispheres, without forming a focal cohesive mass. Widespread infiltration of lymphoma cells in both gray and white matter is characteristic. This condition—also known as lymphomatosis cerebri—is uncommon, occurring in less than 5% of cases, and is a pattern, not a distinct disease entity.
Magnetic Resonance Imaging Brain With and Without Gadolinium
The cornerstone of diagnostic testing, whenever PCNSL is suspected, is MRI with gadolinium. Over three-quarters of DLBCLs in immunocompetent patients are iso- or slightly hypointense compared with gray matter on T1 and iso- or hyperintense on T2. FLAIR signal is also usually iso or hyperintense ( Fig. 84.2A–C ).
On imaging, a single lesion is present in about 60% to 70% of the PCNSL. Keeping in mind that the disease is highly infiltrative and lymphomatous, dissemination in the brain and leptomeninges may occur in the absence of contrast enhancement on magnetic resonance imaging (MRI). Multifocal disease is observed in a substantial proportion of patients, especially at relapse, and is somewhat more common in immunocompromised individuals. Moreover, in immunocompromised patients, the enhancement may be less robust, or completely lacking ( Fig. 84.3A,B ). About 60% of PCNSLs involve the supratentorial space, including the frontal (15%), temporal (8%), parietal (7%), and occipital (3%) lobes, basal ganglia, periventricular brain parenchyma (10%), and corpus callosum (5%). Less frequently involved sites include the posterior fossa (13%) and spinal cord (1%). Furthermore, diffuse enhancing supra and/or infra tentorial pachymeninges and the extension along the (sub) ependymal surfaces can be seen with contrast-enhanced MRI in about 3% to 5% of patients, which is indicative of leptomeningeal infiltration. Because of their high cellularity, over 95% of PCNSLs show mild to moderate diffusion-restriction with apparent diffusion coefficient (ADC) and diffusion weighted imaging (DWI) sequences in a non-vascular distribution.
Magnetic Resonance Spectroscopy
The findings on magnetic resonance spectroscopy (MRS) are nonspecific and include increased choline (a marker of cell membrane transition and proliferation of cells) and lipid (a marker of tumor necrosis) peaks, a moderately decreased N-acetyl-L-aspartic acid peak (a marker of mature neurons), and a slightly decreased creatine peak (negatively associated with energy metabolism). There is interest in development of novel, advanced imaging approaches to detect and provide prognostic information on PCNSL more precisely.
Neuroimaging and Differential Diagnosis Prior to Biopsy
Brain imaging, with single or multiple enhancing lesion(s), raises a broad differential diagnosis other than PCNSL which includes infectious, vascular, autoimmune, and other nonlymphoid neoplastic entities. The comprehensive clinical history as well as several specific radiographic features can assist with contraction of the differential diagnoses. Even when a nonneoplastic etiology resides atop the differential diagnosis, a stereotactic biopsy may have substantial value in clarifying the diagnosis.
Stroke
Certain acute and subacute vascular events can often be confused with CNS lymphoma. Acute/subacute strokes can be enhanced on postcontrast imaging due to BBB breakdown, particularly in the subacute timeframe. Typically, this enhancement is not homogenous and usually dissipates over time. Such contrast enhancement and associated restricted diffusion follow a vascular distribution with a wedge-like appearance in medium to large vessel distribution, or more punctate enhancement in small vessel distribution. As strokes can be in multiple distributions, particularly when secondary to a cardioembolic or vasculitis source, it is not surprising that their radiographic appearance can sometimes be conflated with CNS lymphoma. The clinical pace of symptom(s) in the setting of stroke is typically sudden; however, that of CNS lymphoma is usually subacute but this by itself does not exclude the possibility of underlying stroke (see Chapter 144 ).
Infection
Several CNS infections, including viral, bacterial, mycobacterial, fungal, and parasitic etiologies, are associated with contrast enhancement. While the radiographic appearance does not clearly delineate an infectious etiology, the clinical presentation can provide cues to an underlying, active infection.
Immune Mediated Diseases
Certain autoimmune entities can present similarly as CNS lymphomas with regard to their subacute presentation and brain enhancement on MRI scan. For example, multiple sclerosis (MS) can mimic CNS lymphoma. However, as opposed to PCNSL in immunocompetent hosts, the median age of diagnosis is earlier in MS. Also, MS rarely presents with multiple enhancing brain lesions. Unlike PCNSL, brain enhancement is in a relatively small area in MS. The classic enhancement pattern noted in MS is an “open C” with an incomplete ring of enhancement and the open portion facing medially. Other entities that can mimic CNS lymphoma include acute disseminated encephalomyelitis (ADEM) and neurosarcoidosis. ADEM is often monophasic and can present with single, or multiple, large areas of brain enhancement. Sarcoidosis can often be a chronic process, usually with systemic involvement, along with acute/subacute exacerbations. However, even in these settings, it is often necessary to obtain CNS tissue for pathological diagnosis. All of the autoimmune entities are less likely to exhibit restricted diffusion on ADC/DWI sequences.
Other Central Nervous System Cancers
Infiltrating gliomas may have a similar appearance to CNS lymphoma. Both tumor types can involve the corpus callosum with thickening of the anatomic structure as opposed to the atrophy seen in MS. CNS lymphoma often involves the deeper nuclear supratentorial structures as opposed to infiltrating gliomas, which often affect cortical to subcortical brain tissue. An exception to this rule is gliomas in patients who harbor the H3K27 mutation, which is associated with involvement of deeper midline structures. Low-grade infiltrating gliomas often do not typically enhance. The enhancement pattern of high-grade gliomas (including glioblastoma) is usually heterogeneous, in contrast to the more homogenous pattern of enhancement in CNS lymphoma. In addition, the pattern of restricted diffusion on DWI and ADC sequences is often heterogeneous, when present at all, in high-grade gliomas as opposed to being homogenous in CNS lymphoma.
Tissue Diagnosis
Brain Stereotactic Biopsy
The diagnostic procedure of choice, with the least risk of morbidity and mortality, remains stereotactic biopsy of the brain lesion. If ocular (vitreous, and retina) lymphoma is suspected, then surgical intervention with timely (tissue procurement to laboratory testing) fluid assessment is absolutely essential. Of importance, unnecessary tests should be avoided in order not to jeopardize patient care. For example, analysis of CSF should not delay stereotactic brain biopsy in a non-HIV individual, and CSF fluid testing should be done promptly in cases of suspected isolated PVRL (discussed in detail later).
Misinterpretation of Stereotactic Biopsy
Administration of steroids prior to procurement of brain tissue may significantly decrease enhancement and the size of the tumor, which in turn can easily alter the diagnosis. It is important to inform the neuropathologist of prior systemic steroid exposure to avoid misinterpretation of specimen as “encephalitis”. Biopsy should be repeated in cases where initial biopsy is inconclusive, unsatisfactory, or the diagnosis does not align with the clinical and/or radiologic presentation. In selected cases, a distinct tumor mass is not evident on imaging and/or the biopsy may have been taken from the periphery of the tumor, resulting in the finding of an entirely interstitial pattern with isolated tumor cells intermingled between astrocytes. Sometimes cases with scant tissue may show a relatively monomorphic cell population with intermingled macrophages, mimicking Burkitt lymphoma (see Chapter 85 ).
Risks With Brain Biopsy
Most brain biopsy procedures are safely performed, but the risk of complications is about 8.5%, consisting of hemorrhage, seizures, or brain edema. Biopsy-related mortality is estimated to be 0.9%. Of note, the risk of hemorrhage is 11.5% in individuals with AIDS, compared to only 1% in clinically stable immunocompetent individuals. For this reason, in HIV-infected individuals stereotactic biopsy may be utilized once noninvasive tests are inconclusive or undiagnostic (discussed later). Current evidence does not support diagnostic or therapeutic surgical resection unless there is an impending brain herniation, or rapidly deteriorating neurologic function, failing best conservative practices. Work by some investigators supports more extensive resection. However, such an approach remains non-routine.
Cerebrospinal Fluid Assessment for Diagnosis
Under most circumstances, cerebrospinal fluid (CSF) analysis is no longer required in patients suspected of PCNSL. The CSF findings in PCNSL, if performed, are nonspecific and include elevated protein and lymphocytic predominant pleocytosis with normal glucose levels. A routine CSF examination can only provide definitive evidence of PCNSL in the presence of extensive leptomeningeal dissemination ( Fig. 84.4A–D ). 95,96.109 Importantly, a misinterpretation of a peak caused by a single B cell as a monoclonal population (“pseudomonoclonality”) may result in a false-positive PCR. False positive results can sometimes be minimized by repeating the test and/or through repeated PCR cycles in independent runs. Positive molecular findings in CSF should be interpreted in the context of clinical and tissue morphological data.
Cerebrospinal Fluid Testing in Human Immunodeficiency Virus–Infected Individuals
In selected cases where diagnostic brain biopsy is contraindicated and routine CSF cytology and cytometry are non-diagnostic, special tests may facilitate PCNSL diagnosis. Antinori et al. reported that the combination of increased uptake on thallium-201 SPECT and a positive EBV by PCR in the CSF had 100% sensitivity and 100% negative predictive value. They suggested that patients with positive results could go directly to therapy, obviating the need for tissue biopsy. These findings are provocative and require further validation as few, if any, diagnostic studies have perfect sensitivity and specificity. In another study, the sensitivity and specificity of PCR for EBV-DNA detection in CSF were 80% (95% CI: 60.9% to 91.6%) and 100% (95% CI: 92.6% to 100%), respectively.
Ophthalmology Exam and Vitreoretinal Specimen Testing for Diagnosis
In an appropriate clinical context, a comprehensive ophthalmology exam using dilated fundoscopy, fluorescent angiography, and/or optical coherent tomography is sufficient (without tissue procurement) to diagnose (primary or secondary) VRL, sparing the patient a more invasive vitreous biopsy. For example, the diagnosis of VRL can be established in an older patient with complaints of floaters, vitreous and/or subretinal cellular infiltration on exam, and positive CD20+ large cell cytology and/or flow cytometry for monoclonality in the CSF ( Fig. 84.5A–C ) . However, in a patient with non-diagnostic CSF (as in the majority of cases) and/or brain biopsy, the gold standard for the diagnosis of VRL remains histologic identification of lymphoma cells in the vitreous and/or retina. The tissue diagnosis would require an invasive procedure, such as vitrectomy or vitreous aspirate, retinal or chorioretinal biopsy, or, rarely, diagnostic enucleation.
The reported rates of sensitivity and specificity of cytology for the diagnosis of VRL vary widely, but cytology alone is able to confirm VRL in 45 to 60% of cases. Alternatively, multicolor flow cytometry (MFC) has been successfully employed for phenotyping of vitreal aspirates, with a reported sensitivity of 82% and 100% specificity. The sensitivity of clonality studies ranges between 65% and 95%, depending on the choice of primer sets and quality of material. The limitations of molecular testing, particularly with the small sample size from the vitreous and/or retina, are that results could be false-positive or false-negative. In order to avoid misdiagnosis of minor clonal expansions as evidence for lymphoma, all tests should be run in duplicate to confirm the presence of a dominant clone, and results should be interpreted with caution and only in the context of clinical and morphological findings. Therefore, microdissection identifying the area with most atypical lymphoid cells is critical, and if the patient has been placed on systemic or local corticosteroids, it is best to wait a few weeks to allow more of a vitreal infiltration before doing a tissue biopsy. If the biopsy is nondiagnostic and there is still strong suspicion of VRL, then the other vitreous should be biopsied; or if there are aggregates of subretinal cells, a retinal biopsy using the same vitrector technique with a lower cut rate and manual aspiration should be done. Additional adjuncts in diagnosing VRL exist, including elevation of interleukin-10 (IL-10) or IL-10:IL-6 ratio greater than 1 in ocular fluids, detection of IgH , as well as evaluation of mutant L265P MYD88 (positive in 60% to 80% of cases) and intraocular concentrations of the CD79B Y196 mutation. In cases of mutant L265P MYD88 in the context of CD20+cells in the vitreous, a diagnosis of VRL can be confirmed.
Histology Characterization
Most CNS lymphomas are of B-cell origin, predominantly DLBCL. However, Burkitt lymphoma, acute lymphoblastic lymphoma, marginal zone lymphoma, extraosseous plasmacytoma, and intravascular B-cell lymphoma are rarely encountered. Moreover, less than 4% of PCNSL are of T-cell origin. The morphology cannot discern between PCNSL and SCNSL; it is the role of staging to do so. Microscopically, the neoplastic cells are of medium to large size, with round, oval, irregular, or pleomorphic nuclei and distinct nucleoli, corresponding to centroblasts or immunoblasts, and grow in a characteristic perivascular pattern, forming concentric rings (“angiocentric” or “cuffing”) ( Fig. 84.6A–D ).
JAK/STAT signaling pathway activators such as IL-4 and IL-10 were found to be upregulated and create a favorable TME as evident by characteristic angiotropism. Rubenstein et al. showed that a high expression of activated STAT6 in tumors was associated with short survival in an independent set of patients with PCNSL who were treated with HD-MTX therapy. Infiltration of cerebral blood vessels causes fragmentation of the argyrophilic fiber network. from these perivascular cuffs, tumor cells invade the neural parenchyma. The TME is often accompanied by prominent microglial activation and a reactive inflammatory infiltrate, which consists of mature T cells and, sometimes, lipid-laden (foamy) histiocytes. Notably, in a study by Ponzoni et al. , perivascular T cell infiltrates in PCNSL may predict a favorable outcome, suggesting that immunotherapeutic interventions that potentiate T-cell-mediated immune surveillance may be effective. Centrally, large areas of geographical necrosis are not uncommon making the histologic diagnosis sometimes cumbersome. PCNSL does not have the degree of vascular proliferation commonly observed in glioblastoma. In the immunocompromised host, however, histologic findings of necrosis and hemorrhage are common.
The cells of PCNSL are mature B cells, with PAX5+, CD19+, CD20+, CD22+, and CD79a+ phenotypes ( Fig. 84.6D ). IgM and IgD, but not IgG, are expressed, with either kappa or lambda light chain restriction. Most cases express BCL6 (60% to 80%) as well as MUM1 (90% to 100%), whereas plasma cell markers (i.e., CD38 and CD138) are usually not expressed. Expression of CD10 is unusual in PCNSL (<10%), and its positivity should prompt a search to exclude disseminated systemic DLBCL. Most cases (~80%) of PCNSL of the CNS are positive for BCL2 and MYC expression. However, MYC translocations are rare and BCL2 translocations are absent, suggesting that alternate mechanisms drive the protein expression in PCNSL Conversely, BCL6 protein expression is strongly correlated with the presence of BCL6 rearrangements. The Ki-67 is often markedly elevated; most lesions have a proliferative index that exceeds 70%. The presence of EBV should prompt evaluation for an underlying immunodeficiency. The V600E BRAF protein expression has not been detected in PCNSL.
Biomarkers for Diagnosis
Numerous biomarkers in the serum, CSF, and vitreous fluid have been evaluated with limited success in an effort to possibly obviate the need for invasive diagnostic procedures and to expedite the diagnostic process. Below are data of selected diagnostic (and in some cases prognostic) biomarkers that may be useful in clinical scenarios where diagnosis may be challenging.
Serologic Biomarkers
-
Cell-free DNA (cfDNA; liquid biopsy): In a study by Fontanilles et al., serum cfDNA and matched tumor DNA (tDNA) from 25 PCNSL patients were sequenced. First, patient-specific targeted sequencing of identified somatic mutations in tDNA was performed. Then, a second sequencing targeting MYD88 c.T778C was performed and compared to serologic samples from 25 age-matched control patients suffering from other types of cancer. According to the patient-specific targeted sequencing, 8 patients (32% [95% CI 15 to 54%]) had detectable somatic mutations in cfDNA. Considering MYD88 sequencing, 6 patients had the specific c.T778C alteration detected. Using a control group, the sensitivity was 24% and the specificity was 100%.
-
miR-21 levels: Mao et al. reported that the serum miR-21 was significantly increased in patients with PCNSL when compared with other brain tumors and normal controls in both the test and validation cohort.
Cerebrospinal Biomarkers
-
Circulating cell-free tumor DNA (ctDNA): In the study by Grommes et al., of 9 relapsed/refractory PCNSL patients, tumor-specific ctDNA was found in the CSF of all patients. Sustained tumor responses were associated with clearance of ctDNA from the CSF.
-
IL-10: Meta-analyses by Wang et al. reported a sensitivity and a specificity of IL-10 in CSF for diagnosing CNSLs of 81% (95% CI: 66% to 91%) and 97% (95% CI: 83% to 100%), respectively. Moreover, elevated IL-10 was associated with shorter PFS (hazard ratio: 2.89, 95% CI: 1.13 to 7.41, P = .027).
-
IL-10 and IL-10/IL-6 ratio: Song et al. reported a diagnostic sensitivity and a specificity of 95.5% and 96.1%, respectively (AUC, 0.957; 95% CI, 0.901 to 1.000) using an IL-10 cutoff value of 8.2 pg/mL in the CSF. In addition, the sensitivity and specificity of 95.5% and 100.0% (AUC, 0.976; 95% CI, 0.929 to 1.000), respectively, were reported using an IL-10/IL-6 cutoff value of 0.72. An increased CSF IL-10 level at diagnosis and post-treatment was associated with poor PFS for patients with PCNSL ( P = .0181 and P = .0002, respectively).
-
IL-10 and CXCL-13: In the multicenter study by Rubenstein et al. ( n = 220), a bivariate elevation of IL-10 plus CXCL13 was 99.3% specific for primary and secondary CNS lymphomas, with a sensitivity two-fold greater than cytology or flow cytometry.
-
CXCL13 and CXCL9: Masouris et al. reported a sensitivity and a specificity of about 90% for the diagnosis of active CNSL using a cut-off value of 80 pg/mL for CXCL13. The sensitivity and specificity of CXCL9 were 61.5% and 87.1%, respectively, using a cut-off value of 84 pg/mL.
-
Soluble CD27 (sCD27): Murase et al. reported that sCD27 greater than 15 U/ml was associated with PCNSL, and inflammatory neurological diseases and not with other neurologic diseases or brain tumors.
-
Osteopontin: This is a proinflammatory cytokine involved in immune cell activation and B-cell migration and proliferation. Strehlow et al. reported significantly higher osteopontin levels (>620 ng/mL) in CNSLs compared to patients with inflammatory conditions and GBM or in healthy controls ( P < .01). Multivariate analysis reported that high osteopontin levels were associated with shorter PFS (HR 1.61, 95 % CI 1.13 to 2.31; P = .009) and OS (HR 1.52, 95 % CI 1.04 to 2.21; P = .029).
-
Neopterin: This is regarded as a good biomarker of immune activation mediated by the “lymphocyte-macrophage axis.” Viaccoz et al. reported a sensitivity and a specificity of 96% and 93%, respectively, for the diagnosis of PCNSL at a neopterin cut-off value of 10 nmol/L. The positive and negative predictive values were 84% and 98%, respectively. Geng et al. reported higher neopterin levels in PCNSL compared with other brain tumors and inflammatory conditions.
-
miR-21, miR-19, and miR-92a: Baraniskin et al. reported increased levels of miR-21, miR-19, and miR-92a in PCNSL. Receiver-operating characteristic analyses showed AUCs of 0.94, 0.98, and 0.97 for miR-21, miR-19, and miR-92a, respectively, in discriminating PCNSL from inflammatory and other neurologic conditions.
-
Proliferation-inducing ligand (APRIL) and B-cell activating factor (BAFF): Mulazzani et al. reported that high levels of APRIL and BAFF reliably differentiated CNSL from other focal brain lesions (including primary and metastatic brain tumors, autoimmune-inflammatory lesions, and neuroinfectious lesions) with a sensitivity of 62.3% (APRIL), 47.1% (BAFF), and a specificity of 93.7% (APRIL, BAFF).
-
Multimarker algorithm: Maeyama et al. reported a multimarker diagnostic model using CSF CXCL13, IL-10, β2-MG, and sIL-2R from the results of the case-control study and then applied the model to a prospective study ( n = 104) to evaluate its utility. The multi-marker diagnostic algorithms had excellent diagnostic performance: the sensitivity, specificity, positive predictive value, and negative predictive value were 97%, 97%, 94%, and 99%, respectively.
Vitreous Biomarker for Vitreoretinal Lymphoma
-
miR-155: Tuo et al. reported significantly higher levels of miRNA-155 in uveitis compared to lymphoma.
Future studies with better understanding are required to validate and standardize the diagnostic utility of these and other diagnostic biomarkers in primary and secondary CNSLs.
Staging
The goal of staging in CNSL is to decipher (i) between PCNSL and SCNSL and (ii) in cases of PCNSL, to determine the extent of the disease within the neuraxis. Both of these goals can be accomplished with the tests displayed in Table 84.3 , some of which are guided by individual’s age, immune status, and specific signs and symptom(s), while others are performed routinely. For restaging, all the positive tests, including an ophthalmology exam, are repeated at a predetermined interval (e.g., induction and consolidation); response to therapy is categorized according to International PCNSL Collaborative Group (IPGLS) criteria (described later). Cognitive testing may be performed prior to, during, and after therapy. Correa et al. proposed standardization of neuropsychological evaluation in PCNSL, which includes a battery of tests and a questionnaire to assess domains sensitive to disease and treatment effects such as attention, executive functions, memory, psychomotor speed, quality of life, and premorbid IQ.
Table 84.3
Staging of Primary Diffuse Large B-Cell Lymphoma of the Central Nervous System
| All patients |
|
| Special considerations in immunocompromised individual |
|
CNS , Central nervous system; CSF , cerebrospinal fluid; CT , computed tomography; EBV , Epstein-Barr virus; HIV , human immunodeficiency virus; MRI , magnetic resonance imaging; PCR , polymerase chain reaction; PET , positron emission tomography; PTLD , post-transplant lymphoproliferative disease.
Frontline Therapy and Response Assessment
Immediate Conservative Management: Best Practices
PCNSL patients who present with focal brain lesion(s) are often symptomatic at diagnosis, with an emergent to urgent requirement for therapy. Based on MRI findings, many clinicians preemptively start steroids. Priority should be given to immediate nonsteroid-based conservative management and neurosurgical consultation for immediate stereotactic biopsy. Prediagnosis steroids usually lyse the neoplastic B cells, leaving behind reactive cells that may cause false-negative interpretation of biopsy as “encephalitis”, a phenomenon often referred to as “vanishing tumor”. In the study by Brück et al., corticosteroids prevented diagnosis in as many as 50% of the patients. Therefore with the rare exception of impending or frank herniation, the use of prediagnostic steroids is strongly discouraged. Priority should be given to mannitol or hypertonic saline over steroids. There is no standard dosing of steroids for patients who must receive them. Also, there is no role for prophylactic anti-seizure therapy.
Induction Therapy
The treatment for patients who are deemed “medically fit” should be initiated with a HD-MTX-based regimen with the goal of achieving a complete radiologic response (CR or unconfirmed CR [CRu]), defined as a complete resolution of all enhancing diseases on the brain MRI, in the absence of corticosteroid use, for at least 2 weeks before imaging. If the disease was initially noted in other CNS compartments, CR is similarly defined by the absence of disease in previously positive space(s). The consensus criteria for disease response were established by the International PCNSL Collaborative Group (IPCG), and defined as follows: CR, CRu, partial response (PR), stable disease (SD), or progressive disease (PD). A CRu may still demonstrate radiographic abnormalities of uncertain significance, precluding the imaging from qualifying as a CR. Often, CRu findings are secondary to the initial diagnostic biopsy.
Commonly used induction regimens are displayed in Table 84.4 . They include: HD-MTX/vincristine/procarbazine/rituximab ( MVP-R ; CR of 78% [Cl not provided]); HD-MTX/temozolomide/rituximab ( MTR ; CR rate of 66% [CI, 50% to 80%]); rituximab/HD-MTX/thiotepa/cytarabine ( MATRix ; CR rate of 49% [95% CI 38 to 60]); and HD-MTX/BCNU/teniposide/prednisone/rituximab ( MBTP ). Of high importance, the difference in induction CR rates must be interpreted in the context of the details of study design, number of patients, and specific toxicities related to each regimen. (See box on Treatment of Newly Diagnosed Adult Patient With Primary Central Nervous System Lymphoma .)
Table 84.4
Commonly Utilized High-Dose Methotrexate-Based Induction Regimens in Diffuse Large B-Cell Lymphoma of the Central Nervous System
| Induction regimen | Schedule and Doses |
|---|---|
| MVP-R N = 30 |
|
| MTR (Alliance 50202) N = 43 |
|
| MATRix (IELSG32) N = 75 |
|
| MBTP (HOVON 105/ALLG NHL 24) N = 100 |
|
CR , Complete remission; CSF , cerebrospinal fluid; HD-MTX , high-dose methotrexate; IT MTX , intrathecal methotrexate; IT , intrathecal; IV , intravenous; N , patients.
You're Reading a Preview
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here