Abstract
Background/Aim: Primary central nervous system diffuse large B-cell lymphoma (CNS DLBCL) is a rare entity, accounting for 3-4% of intracranial neoplasms. This study aimed to investigate the clinicopathological characteristics of primary CNS DLBCL patients and their prognostic implication. Patients and Methods: We collected 74 cases of clinically and pathologically confirmed primary CNS DLBCL from two institutions. Disease-free survival (DFS) and overall survival (OS) were analyzed based on various clinicopathological parameters. Results: Most cases (83.8%) were classified as activated B-cell immunophenotype by Hans algorithm and cell-of-origin classification did not influence the clinical outcome. On univariate analysis, age (>60 years) and ECOG performance status (≥2) were significantly associated with shorter DFS and OS, and MYC/BCL2 co-expression significantly impacted poor DFS. An anaplastic variant was diagnosed in only 2 cases, but it raised possible association with poor outcome. On multivariate analysis, ECOG performance status and age was associated with poor prognosis. Conclusion: In primary CNS DLBCL, age and performance status revealed the most significant association with prognosis. Cell-of-origin classification was not a significant prognostic factor in contrast to systemic DLBCL.
Primary diffuse large B-cell lymphoma of the central nervous system (CNS DLBCL), formerly named primary CNS lymphoma, is a primary DLBCL confined to the CNS, including the brain, spinal cord, leptomeninges, and eyes. CNS DLBCL accounts for approximately 3-4% of intracranial neoplasms (1).
Systemic DLBCLs have a variety of morphologic and genetic spectra associated with variable clinical outcomes (2). Gene expression profiling (GEP) based on DNA microarray platform identified two major groups of systemic DLBCLs corresponding to different stages of B-cell differentiation [cell-of-origin (COO) classification] (3). The first group expresses signature genes of germinal center B-cells (GCB) – type, and the second group expresses genes normally incorporated during in vitro activation of peripheral blood B-cells – activated B-cell (ABC) type. The GCB, ABC, and unclassifiable types account for approximately 50%, 30%-40%, and 15%-20% of all systemic DLBCLs, respectively (4). The COO classification distinguishes significantly different clinical outcomes, as well as fundamental biological differences. The GCB group was shown to have a much better 5-year overall survival than the ABC group: 76% versus 16% in the original study (3), and in the rituximab era, the survival difference was retained as 80% versus 50% (5).
Double-hit lymphoma (DHL) is characterized by high-grade B-cell lymphoma with translocation of both MYC and B-cell lymphoma (BCL2), and double-expressing lymphoma (DEL) is defined as DLBCL with co-expression of MYC and BCL2 on immunohistochemical staining (6). DEL is found in approximately 30% of all DLBCL cases and is associated with aggressive clinical behavior. More recently, several molecular classification systems have been evolved, based on different sequencing and computational analyses (7-10). There are distinct genetic subtypes of DLBCL across the different molecular classifiers. One of these was characterized by MYD88 and CD79B mutations (MCD) and strongly associated with ABC DLBCLs and the extranodal lymphomas, including primary CNS DLBCL, primary breast lymphoma, and primary testicular lymphoma (11). Due to its rarity, more data on primary CNS DLBCL are required to be comprehensively accumulated.
In this study, we investigated the clinicopathological characteristics of the primary CNS DLBCL patients of two different tertiary referral hospitals. We collected a total of 74 patients with primary CNS DLBCL and evaluated various aspects of the clinical and pathological features. Furthermore, we analyzed the clinical impact of the features in the primary CNS DLBCL patients.
Patients and Methods
Patients. The series included 74 cases of DLBCL arising in the brain from two institutions [Kyung Hee University Hospital (Seoul, Republic of Korea) and Kangbuk Samsung Hospital (Seoul, Republic of Korea)]. All cases were pathologically confirmed by biopsy. The cases were consecutively collected from January 2010 to January 2020. The exclusion criteria were: lymphomas of the skull or dura; intravascular large B-cell lymphomas; lymphomas with evidence of systemic disease or secondary lymphomas; and Epstein-Barr virus (EBV) or human immunodeficiency virus infection (2). This study was approved by the Institutional Review Board of the Kyung Hee University Hospital (KHUH 2021-06-036-004).
Review of clinicopathological findings. We collected all available clinical data including age, sex, tumor location, number of tumors, presence of other CNS disease, pre-operative serum level of lactate dehydrogenase (LDH), Eastern Cooperative Oncology Group (ECOG) performance status, treatment, and disease progression status. The primary endpoints were defined as follows: Overall survival (OS) was defined by the interval between diagnosis and death regardless of cause; and disease-free survival (DFS) was defined as the time from diagnosis to recurrence or death whichever came first.
Two experienced pathologists (KY.N. and H.Y.W.) reviewed hematoxylin and eosin slides of 45 cases with routine microscopic examination. The histological variant was classified as: centroblastic variant (medium-sized to large monomorphic lymphoid cells with 2-4 nucleoli and scant cytoplasm); immunoblastic variant (lymphoid cells with a single nucleolus and an amount of basophilic cytoplasm, often with plasmacytoid differentiation); and anaplastic variant (large to very large lymphoid cells with bizarre pleomorphic nuclei resembling Reed-Sternberg cells).
Immunohistochemistry (IHC). All cases were stained with cluster of differentiation 20 (CD20), CD3, B-cell lymphoma 2 (BCL2), BCL6, CD10, MYC, multiple myeloma 1 (MUM1), and EBV in situ hybridization. MYC and BCL2 immunostaining were available in 58 and 69 cases, respectively, due to tissue loss during tissue section. Immunostaining was performed on a whole block tissue using an automatic instrument (Vision Biosystems, Mount Waverly, Victoria, Australia) according to the manufacturer’s recommendations. An antigen retrieval procedure was performed using the Bond ER2 Solution for 30 min at 100°C. Endogenous peroxidases were quenched by incubation with hydrogen peroxide for 5 min. The sections were incubated for 15 min at ambient temperature with primary antibodies against CD20 (1:2000, clone L26, Leica Biosystems, Newcastle, UK), CD3 (1:200, clone polyclonal, DAKO, Glostrup, Denmark), BCL2 (1:50, clone bcl-2/100/D5, Leica Biosystems), BCL6 (1:1, clone LN22, Leica Biosystems), CD10 (1:100, clone 56C6, Leica Biosystems), and MYC (1:100, clone EP121, Cell Marque, Rocklin, CA, USA), MUM1 (1:100, clone MUM1p, DAKO). Nuclei were counterstained with hematoxylin. A negative control was prepared by substituting non-immune serum for primary antibody, which resulted in no detectable staining.
In situ hybridization. To determine EBV infection status of the lymphoma cells, in situ hybridization for EBV-encoded RNA (EBER; Leica Biosystems, Newcastle, UK) was performed according to the manufacturer’s instructions on an automated Leica Biosystems. Staining in ≥80% of lymphoma cells was considered positive.
Assessment of immunohistochemistry. The expression of the IHC markers was estimated in 10% increments. Two independent pathologists (SI.D. and KY.N.), who were blinded to the clinicopathological data and the patients’ identities, participated in the immunohistochemical scoring. COO was determined based on Hans algorithm (12), by estimating the expression of CD10 (membrane, ≥30% positive), MUM1 (nucleus, ≥30% positive), and BCL6 (nucleus, ≥30% positive). The GCB subtype had a CD10+ or CD10−, BCL6+ and MUM1− phenotype, whereas CD10− and BCL6− or CD10−, BCL6+, and MUM1+ phenotype was assigned as ABC subtype. For additional IHC, MYC was considered positive if ≥40% of tumor nuclei are positive, and BCL2 was considered positive if ≥50% of tumor nuclei are positive, as previously described (13, 14).
Statistical analysis. Chi-square test or Fisher’s exact test were performed to compare the means between non-continuous variables. Kaplan–Meier survival curves and log-rank statistics were used to evaluate OS and DFS. Multivariate regression analysis was performed using the Cox proportional hazards model (95% confidence interval) with the backward stepwise elimination method. The Cohen’s kappa value was used to compare the agreement of two reviewers (KY.N. and H.Y.W.) in 74 cases. All statistical analyses were performed using IBM SPSS Statistics for Windows, version 22.0 (IBM Corp., Armonk, NY, USA). All statistical tests were two-tailed; p<0.05 was considered statistically significant, and 0.05≤p<0.1 was considered a trend toward significance, to increase the sensitivity to detect potential selection bias.
Results
The clinicopathological characteristics of the primary CNS DLBCL. The baseline clinicopathological characteristics of the primary CNS DLBCL patients are summarized in Table I. Of a total of 74 patients, 46 patients (62.2%) were male and 54 patients (73.0%) were over 60 years of age. Most cases were located in the non-midline site, and the most common site was the cerebral hemisphere (n=32, 43.2%). Twenty-one patients (28.4%) had multiple tumors with intracranial lymphomatosis. Seventeen patients (23.0%) had prior history of other CNS disease, including cerebral infarction (16.2%), neurodegenerative disease (2.7%), CNS neoplasm (2.7%, pituitary adenoma and thalamic low-grade glioma), and intracranial hemorrhage (1.4%). Nineteen patients (25.7%) presented with elevated serum lactate dehydrogenase (LDH) pre-operatively. Eighteen patients (24.3%) had an ECOG status of 2 or higher.
The baseline characteristics of the primary central nervous system (CNS) diffuse large B-cell lymphoma (DLBCL) patients.
Most patients (n=65, 91.5%) received chemotherapy after diagnosis, either alone (n=49, 69.0%) or combined with radiation therapy (n=16, 22.5%). Only a small subset (n=6, 8.4%) underwent radiation therapy alone and 10 patients (14.1%) received no or palliative treatment. Among the 64 patients whom survival data were available, 36 patients (56.3%) died during follow-up [time to death (months), mean±standard deviation (SD); 20.5±22.9], 17 patients (26.6%) recurred during follow-up duration [time to recurrence (months), mean±SD; 36.5±50.7].
Immunohistochemically, only 7 cases (9.5%) had CD10 expression, and the vast majority harbored MUM1 (n=66, 89.2%) and BCL6 (n=64, 86.5%) expression. MYC and BCL2 over-expression were observed in 45 cases (77.6%) and 58 cases (84.1%), respectively. MYC/BCL2 co-expression was noted in 37 cases (63.8%). According to the COO concept of classification, most cases (n=62, 83.8%) were classified as ABC immunophenotype by Hans algorithm. Additionally, all cases were negative for EBV in situ hybridization. Morphologically, the most common cases (n=32, 71.1%) were centroblastic, followed by immunoblastic (n=11, 24.4%), and anaplastic (n=2, 4.4%).
The association of clinicopathological characteristics and COO classification. The various clinicopathological features of the primary CNS DLBCL patients were evaluated according to the COO classification (Table II). There was no sex predilection and no difference in age, location, the number of tumors, serum LDH level, and ECOG performance status (>2) (p>0.05 for all). Unexpectedly, presence of other CNS disease was significantly prevalent in GCB-type primary CNS DLBCL patients (p=0.004). More than half of patients with the GCB-type primary CNS DLBCL, (7/12, 58.3%) experienced also another CNS disease (5 cerebral infarction, 1 thalamic low-grade glioma, and 1 Parkinson disease). On the other hand, only 16.1% (10/62) of patients with the ABC-type primary CNS DLBCL had also another CNS disease (7 cerebral infarction, 1 intracranial hemorrhage, 1 pituitary adenoma, and 1 Alzheimer disease).
The clinicopathological characteristics associated with the cell-of-origin (COO) classification of the primary central nervous system (CNS) diffuse large B-cell lymphoma (DLBCL) patients.
Pathologically, the COO classification did not show any difference in morphological variants, MYC or BCL2 over-expression, and MYC/BCL2 co-expression (p>0.05 for all).
Survival analysis. We analyzed the clinical impact of clinicopathological variables on survival outcomes in patients with primary CNS DLBCL (Table III). An age over 60 years was confirmed as an independent poor prognostic factor in both univariate [HR (95%CI)=6.68 (1.98-22.54), p=0.002] and multivariate [HR (95%CI)=5.69 (1.21-26.74), p=0.028] analyses for OS. For DFS, it was also associated with worse prognosis in univariate [HR (95%CI)=3.49 (1.43-8.53), p=0.006] and a trend toward significance in multivariate [HR (95%CI)=2.84 (0.91-8.89), p=0.073] analyses. The higher ECOG performance status was significantly associated with shorter DFS and OS in both univariate [DFS, HR (95%CI)=4.85 (2.50-9.42), p<0.001; OS, HR (95%CI)=5.17 (2.60-10.42), p<0.001] (Figure 1A and B) and multivariate [DFS, HR (95%CI)=3.71 (1.67-8.28), p=0.001; OS, HR (95%CI)=4.34 (1.84-10.22), p=0.001] analyses. The co-expression of MYC/BCL2 significantly impacted poorer DFS in univariate analysis [HR (95%CI)=3.31 (1.01-10.90), p=0.049] (Figure 1C and D), but not in multivariate analysis [HR (95%CI)=2.11 (0.63-7.08), p=0.216]. The COO classification and MYC or BCL2 over-expression was not related to the clinical outcome of CNS DLBCL (Figure 1E and F). Statistical analysis on morphological variants was limited as only 2 cases of anaplastic variant was included, but anaplastic variant showed much shorter DFS compared to non-anaplastic cases (mean 4.5 months vs. 33.1 months).
The clinical impact of clinicopathological variables on the primary central nervous system (CNS) diffuse large B-cell lymphoma (DLBCL) patients.
Kaplan–Meier curves of primary central nervous system diffuse large B-cell lymphoma patients. Disease-free survival and overall survival according to Eastern Cooperative Oncology Group (ECOG) performance status (A, B), anaplastic variant (C, D), and cell-of-origin classification (E, F). ABC: Activated B-cell; GCB: germinal center B-cell.
Discussion
We collected 74 cases of primary CNS DLBCL from two different tertiary referral hospitals. Most of our cases demonstrated ABC immunophenotype with high rate of MYC and/or BLC2 over-expression. Age and ECOG performance status were confirmed as independent poor prognostic factors. In addition, MYC/BCL2 double expression significantly influenced shorter DFS in univariate analysis.
To date, several studies have suggested various prognostic scoring models of primary CNS DLBCL patients (15-18). The International Extranodal Lymphoma Study Group provided a 5-point scoring index based on age, ECOG performance status, serum LDH level, cerebrospinal fluid (CSF) total protein concentration, and involvement of deep brain structures (15). Nottingham/Barcelona model designed a 4-point scoring system using age, ECOG performance status, and multifocality (16). And the Memorial Sloan-Kettering Cancer Center proposed a simple but robust model using only age and performance status (17). Asan Medical Center assigned a 3-point scoring system based on age, multifocality, and high CSF protein concentration (18). Of these, the most consistent factors are age and performance status, which were both confirmed as independent prognostic factors in our study as well.
The COO classification based on GEP is regarded as a strong prognostic factor in systematic DLBCL, and the COO subtypes have different responses to chemotherapeutic agents (5, 19). In this study, 74 patients were classified into two subgroups by Hans algorithm, which was found to have high concordance with the microarray results and considered as a useful tool for determination of the COO as defined by GEP (20). However, there are consistent reports showing that the application of the simplified COO classification using IHC does not predict the prognosis in primary CNS DLBCL (21-23), which is in line with the results in our series. Only a single study based on relatively small number cases (n=24) demonstrated that GCB type is a favorable prognostic factor in primary CNS DLBCL patients (24).
DHL and DEL involving gene rearrangement and protein expression of MYC and BCL2 have recently become the most widely used terms. In systemic DLBCL, DEL accounts for approximately 30% (25), and predicts inferior survival in most studies (26, 27). In contrast, in terms of the prognostic value of MYC/BCL2 co-expression in primary CNS DLBCL, a consistent result has not yet been drawn (22, 23, 28). In our case, MYC/BCL2 co- expression was noted in 64% of the cases and significantly associated with poor DFS but failed to remain in multivariate analysis combined with age and performance status.
Anaplastic DLBCL is an uncommon morphologic variant, which accounts for 4.4% of our cases. As for systemic DLBCL, it has been reported that the anaplastic variant is associated with poor prognosis, high International Prognostic Index score and ABC immunophenotype (29). Although statistical analysis was limited due to the small number of anaplastic variant cases in our cohort, they showed significantly shorter DFS compared to non-anaplastic cases, suggesting possible association with poor prognosis.
One of our unexpected results was that history of CNS disease such as infarction, hemorrhage or neurodegenerative disease was significantly more frequently present prior to lymphoma, in the GCB-type than the ABC-type. Given that these diseases are increased with aging, the fact that the mean age of patients with GCB-type is higher than that of the ABC-type (73 vs. 65) in our cohort may explain the reason.
In summary, we performed clinicopathological evaluation and prognostic analysis on 74 primary CNS DLBCL patients treated in two different tertiary referral hospitals. We validated that age and performance status are strong prognostic factors and found that some pathological features may have prognostic implication.
Acknowledgements
This research was funded by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Science, ICT, and Future Planning (NRF-2020R1G1A1003692).
Footnotes
Authors’ Contributions
Conceptualization, SI.D. and KY.N.; data curation, H.Y.W. and KY.N.; writing – original draft preparation, H.Y.W. and KY.N.; writing – review and editing, H.Y.W. and KY.N.; visualization, H.Y.W.; supervision, S.W. C.; project administration, SI.D. and KY.N. All Authors have read and agreed to the published version of the manuscript.
Conflicts of Interest
The Authors declare no conflicts of interest in relation to this study.
- Received August 12, 2022.
- Revision received August 25, 2022.
- Accepted August 26, 2022.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
References
In this issue
Jump to section
Related Articles
Cited By...
- Next Generation Sequencing Analysis of Fibrin-associated Large B-cell Lymphoma Reveals Pathogenic Single Nucleotide Variants
- Gastrointestinal Follicular Lymphoma: A Single-institutional Experience of 22 Cases With Emphasis on the Comprehensive Clinicopathological Analysis and Diagnostic Re-classification