Abstract
Background/Aim: Bruton’s tyrosine kinase (BTK)-mediated B-cell-receptor signaling drives lymphomagenesis of diffuse large B-cell lymphoma (DLBCL). We investigated the clinicopathological significance of BTK positivity in DLBCL according to known molecules related to resistance to BTK inhibitors [BCL2 apoptosis regulator (BCL2)/MYC proto-oncogene, bHLH transcription factor (MYC)]. Patients and Methods: We evaluated BTK expression immunohistochemically in 106 DLBCLs considering their BCL2/MYC status. Results: Considering the whole cohort, BTK was expressed in 65.1%, including 70.4% (50/71) of non-germinal center B-cell-like (non-GCB) subtype; BCL2 expression was detected in 60.4%, MYC expression in 15.1%, MYC translocation in 4.2% (4/96) and MYC gain/amplification in 7.6% (8/105). Overall and in the non-GCB cohort, BTK positively correlated with high international prognostic index (both p=0.005) and stage (p=0.006 and p=0.002), and with BCL2 intensity (p=0.005 and p=0.026, respectively); MYC gain/amplification total cohort (p=0.038). Moreover, high risk, defined as co-expression of BTK and either or both BCL2/MYC, independently predicted shorter progression-free survival in patients treated with rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone (R-CHOP) (all R-CHOP-treated patients: hazard ratio=2.565, p=0.044; R-CHOP-treated non-GCB subgroup: HR=3.833, p=0.019). Conclusion: BTK expression may be utilized to stratify risk in patients with DLBCL.
Bruton’s tyrosine kinase (BTK) is a cytoplasmic non-receptor tyrosine kinase (1) which is widely expressed in hematolymphoid tissues and is a key element in B-cell-receptor (BCR) signaling (2). Development of X-linked agammaglobulinemia by BTK gene mutation implicated BTK in B-cell ontogeny (3). Furthermore, BTK maintains the malignant phenotype of B-cell neoplasms (4, 5), providing the rationale for BTK-inhibitor treatment in B-cell malignancies (6).
BTK-mediated BCR signaling contributes to DLBCL pathogenesis (7, 8). Compared to germinal center B-cell like (GCB) subtype DLBCL, the non-GCB subtype exhibits chronic active BCR signaling (8). BCR signaling molecules such as BTK are emerging as new targets in non-GCB DLBCL. In a phase I/II trial of Ibrutinib, a covalent inhibitor of BTK, the overall response rate was 37% in non-GCB DLBCLs (9). Ibrutinib is currently used in chronic lymphocytic leukemia (CLL), Waldenstrom macroglobulinemia, mantle cell lymphoma and marginal zone lymphoma, and is considered in other malignancies including non-GCB DLBCL (10).
Since BTK-inhibitor trials did not consider BTK expression [reviewed in (6)], only limited data are available for BTK expression in B-cell lymphomas. Fernandez-Vega et al. observed a wide range of BTK expression in DLBCL (11), raising the question of the clinicopathologicaI significance of the BTK level in DLBCL.
During treatment, tumor B-cells acquire resistance to BTK inhibitor (12, 13). In some experiments, combination therapy with ibrutinib and an inhibitor of BCL2 apoptosis regulator (BCL2) showed synergistic effects because ibrutinib-resistant cells of non-GCB DLBCL exhibited BCL2 overexpression (14, 15). Similarly, activation of MYC proto-oncogene, bHLH transcription factor (MYC) signaling induced resistance to BTK inhibitor in precancerous B-cells and mantle cell lymphoma cells (16, 17).
Considering that co-expression of BCL2/MYC defined an aggressive subset of DLBCL with active BCR signaling (7, 18, 19), associations between BTK and BCL2/MYC status need to be explored in perspectives of pathogenesis and prognostication.
Here, we investigated the clinicopathological significance of BTK positivity in DLBCL according to BCL2/MYC status.
Patients and Methods
Patients and samples. A total of 106 cases of de novo DLBCL, not otherwise specified, diagnosed between May 2003 and January 2013 at Seoul National University Bundang Hospital were included in this study. By histological review, distinct entities including ‘T-cell/histiocyte-rich large B-cell lymphoma’, ‘DLBCL of the central nervous system’, ‘primary mediastinal large B-cell lymphoma’, ‘gray-zone lymphomas’ and lymphomas associated with Epstein– Barr virus or human immunodeficiency virus were excluded (7). Cell-of-origin was determined by Hans et al.’s algorithm (20). The study was approved by the Institutional Review Board of Seoul National University Bundang Hospital (B-1306-208-301), by which informed consent was waived considering retrospective design and no increase of risk to the study patients.
Immunohistochemical staining and interpretation. Archived hematoxylin and eosin-stained slides were reviewed in each case to confirm the original diagnosis and select the most representative sections. Cores of 2-mm diameter were taken from representative formalin-fixed paraffin-embedded tissue blocks of patients, and tissue microarrays (TMAs) were prepared as previously described (21). TMAs were sectioned at 4-μm-thickness and stained using Benchmark XT and Benchmark ULTRA (Roche Diagnostics, Risch-Rotkreuz, Switzerland) with the following antibodies: Anti-BCL2 (124, 1:50; Dako, Glostrup, Denmark), anti-MYC (Y69, 1:100; Epitomics, Burlingame, CA, USA). BTK was manually stained with D3H5 rabbit monoclonal antibody (#8547, 1:50; Cell Signaling, Danvers, MA, USA). Antigen retrieval was performed using microwaving for 5 minutes (×2) in Tris-EDTA (pH9, 10×) buffer. Primary antibodies were applied for 2 hours at room temperature. REAL™ EnVision™ Detection System (Dako) was used for signal detection.
Immunohistochemistry was interpreted by two pathologists (YBH and JHP). BTK staining (>50% of tumor cells) was graded as none (score 0), weak (score 1), moderate (score 2) and strong staining (score 3) by average intensity in the cytoplasm and membrane, where 3 was equivalent to BTK expression of reactive mantle B-cells. We defined a score of 3 as BTK-positive, and of 0-2 as BTK-negative (Figure 1). For correlation analysis between BTK and BCL2, BCL2 expression was graded similarly by average intensity in >50% of the tumor cells (22). BCL2 positivity was defined as staining with a score of 2-3 (Figure 2A-D). MYC expression was interpreted using 40% as a cutoff criterion (18, 23) (Figure 2E and F).
Fluorescence in situ hybridization for MYC gene. Fluorescence in situ hybridization was performed using Vysis LSI/Myc(8q24.12-q24.13) SpectrumOrange and Vysis centromere enumeration probe 8 (CEP8)(D8Z2) SpectrumGreen probes (Abbott Molecular, Des Plaines, IL, USA) for copy number gain (CNG; MYC gene/CEP8 ratio>1.5) and amplification (MYC gene/CEP8 ratio>4), and Vysis LSI Myc Dual Color Break-Apart Rearrangement probe for translocation (separate signals in >15% of tumor cells) as previously described (21, 24, 25) (Figure 3).
Statistical analysis. Statistical analysis was performed using Statistical Package for the Social Sciences 21.0 (IBM Corp., Armonk, NY, USA). The chi-square test (or Fisher’s exact test) was used to assess associations. In survival analysis, progression-free survival (PFS) was defined as the time interval from the start of treatment to the date of progression including radiologically confirmed progressive disease using positron-emission tomography-computed tomography or computed tomography refractoriness to the first-line therapy in response evaluation after the second cycle or completion of treatment, relapse or death. A Kaplan–Meier analysis was performed to construct survival curves with log-rank test for comparison between curves. In our cohort, the median PFS was 23.3 months (range=0.1-117.1 months) and during the follow-up period, progression was observed in 25 patients (23.6%). A multivariate analysis was performed by Cox-regression modeling. Statistical significance was accepted for p-values of less than 0.05 (two-sided).
Results
Clinicopathological characteristics of total cohort. As shown in Table I, patients with DLBCLs were divided into those with GCB (27/106, 25.5%), non-GCB (71/106, 67.0%) subtypes and unclassifiable cases (8/106, 7.5%). The median age was 60.0 years, age ranging from 18 to 83 years. The majority of the patients had good Eastern Cooperative Oncology Group performance status (ECOG PS) (<2; 90.6%), low international prognostic index (IPI; 66.0%), and involvement of <2 extranodal sites (75.5%). The patients usually lacked B symptoms (79.2%), bone marrow involvement (85.6%; 83/97) or bulky diseases (91.5%). Compared to the GCB subtype, non-GCB DLBCL tended to have female sex, B symptom, elevated serum lactate dehydrogenase (LDH), high IPI (class 3-5), involvement of ≥2 extranodal sites, and high stage (III-IV). Chemotherapy with rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone (R-CHOP) was the predominant regimen (86.8%).
Expression of BTK in normal B-cells and DLBCLs. In the whole cohort, BTK was positive (score 3; equivalent intensity of mantle B-cells) in 65.1%, and negative (score 0-2) in 34.9%, including 19.8% with a score of 2, 10.4% with a score of 1, and 4.7% with no staining. The non-GCB subtype had a trend for frequent BTK positivity (in 70.4%) compared to the GCB subtype (59.3%) without statistical significance (p=0.137).
Characteristics of BCL2/MYC-related molecular markers in DLBCLs. We also evaluated expression of BCL2 and MYC, the proposed mechanisms of resistance to BTK inhibitor in B-cell lymphomas (14-17) (Table I). Overall, BCL2 was positive in 60.4% and MYC was positive in 15.1% of the total cohort. The non-GCB subtype showed a trend for frequent BCL2 expression compared to the GCB subtype (p=0.111), whereas MYC expression had no predilection for subtype. Coexpression of BCL2 and MYC tended to be more frequent in the non-GCB subtype, without statistical significance (p=0.338). MYC translocation occurred exclusively in GCB subtype (p=0.004) but the frequencies of CNG and amplification were not different between GCB and non-GCB subtypes.
Clinicopathological and molecular features of patients with DLBCL according to BTK positivity. To investigate clinicopathological features of BTK-positive and -negative subgroups, we analyzed correlations between clinicopathological variables and BTK (Table II). Remarkably, BTK positivity correlated with high IPI (p=0.005) and high stage (p=0.006), with concordant trends for ECOG PS (≥2), number of extranodal sites (≥2), and elevated LDH to various degrees. These associations were almost equivalently preserved only in the non-GCB subtype (p=0.005 for IPI and p=0.002 for stage) but not in GCB DLBCLs. These findings indicate that BTK-positive DLBCL was clinically characterized by high IPI and high stage, especially in those with non-GCB subtype.
Associations between BTK and BTK-inhibitor resistance-related molecular markers. To analyze interactions between BTK, BCL2 and /MYC, we investigated associations between their expression and translocation and CNG/amplification of MYC gene (Table III). For sensitive detection, we used the graded scores of BTK and BCL2 expression. BTK intensity positively correlated with BCL2 intensity (ρ=0.268, p=0.005) in the whole cohort and the non-GCB subtype (ρ=0.264, p=0.026) but not in the GCB subtype. Moreover, BTK correlated with MYC CNG considering the whole cohort (ρ=0.203, p=0.038), while BTK was not associated with MYC expression, BCL2/MYC coexpression nor MYC gene translocation. Taken together, BTK intensity was found to be positively correlated with BCL2 intensity, suggesting the close relationship of BTK with BCL2, despite a less obvious association with the MYC pathway.
Survival analysis by BTK/BCL2/MYC risk group. We next investigated the clinical outcomes of the whole cohort (n=106) and the R-CHOP-treated subgroup (n=92) according to conventional clinicopathological variables and the studied molecular markers (Table IV and Figure 4). In univariate analysis for PFS, IPI and its individual components, i.e., age, ECOG PS, serum LDH, number of extranodal sites and stage, were significantly associated with PFS (p<0.001 for IPI both for the total cohort and in R-CHOP-treated subgroup). MYC gene translocation also harbored a strong prognostic value (p<0.001 for both groups). In the analysis of expression of BTK and related molecular markers, BTK positivity reflected an inferior outcome (p=0.013 and p=0.020, respectively). Although BCL2 or MYC did not show prognostic significance, coexpression of BCL2 and /MYC was associated with poor clinical outcome in the R-CHOP-treated subgroup (p=0.047).
On the biologic grounds that ibrutinib-induced suppression of neoplastic B-cells via targeting BCR/BTK signaling might be evaded by BCL2 accumulation or MYC activation (15, 16), we hypothesized that concordant activation of BTK and at least one of these compensatory pathways may be crucial for stably maintaining B-cell survival. We designed the BTK/BCL2/MYC risk group, which defines cases with BTK positivity and the expression of one or more resistance-related proteins (BCL2 and MYC) as high-risk, and remaining cases as low-risk (Table V).
In univariate analysis, the BTK/BCL2/MYC risk group showed the most prognostic significance in PFS (p=0.023 in total cohort; p=0.014 in R-CHOP-treated subgroup) compared to the BTK/BCL2 and BTK/MYC coexpression group. Multivariate analysis incorporating prognostically significant and representative variables in univariate analysis, i.e., IPI, MYC gene translocation and BTK/BCL2/MYC risk score, revealed that BTK/BCL2/MYC risk group was independently associated with prognosis in R-CHOP-treated subgroup (p=0.044, HR=2.565) along with IPI and MYC gene translocation.
Survival analysis in the non-GCB cohort by BTK/BCL2/MYC risk group. Considering the clinical implication of BTK positivity in the non-GCB subtype (Table II), we additionally focused on the survival analysis of those with non-GCB subtype (n=71) and R-CHOP-treated non-GCB subtype (n=63) (Table VI and Figure 5). As shown for the whole cohort, IPI, and its individual components, as well as bulky disease were poor prognostic factors in univariate analysis (p<0.001 for IPI both in non-GCB subtype and in R-CHOP-treated non-GCB subtype). BTK positivity was associated with shorter PFS in patients with non-GCB subtype (p=0.041), while the prognostic significance was limited in those with non-GCB subtype treated with R-CHOP (p=0.060). With respect to the multiple marker variable, high risk was observed as predictive for shorter PFS (p=0.023 in non-GCB subtype; p=0.011 in R-CHOP-treated non-GCB subtype).
Multivariate analysis incorporating BTK/BCL2/MYC risk group, IPI and bulky disease revealed that the high-risk group was independently associated with shorter PFS (p=0.030, hazard ratio=3.107 in non-GCB subtype; p=0.019, hazard ratio=3.833 in R-CHOP-treated non-GCB subtype).
Discussion
Herein, we evaluated the clinicopathologicaI features of patients with DLBCL according to BTK positivity and BCL2/MYC status. We observed that i) BTK was strongly expressed in 65.1% of DLBCLs, with slight predilection for those with the non-GCB subtype (70.4%); ii) BTK positivity correlated with stage and IPI as well as BCL2 expression and MYC CNG; iii) BTK positivity predicted shorter PFS, but not independently from IPI; and iv) being in the BTK/BCL2/MYC high-risk group was an IPI-independent poor prognostic factor for PFS in both the cohort overall and in the non-GCB subtype.
In a previous study by Fernandez-Vega et al., BTK was expressed in 86% (94/109) of DLBCLs with various intensities (strong in 6%, moderate in 48%, weak in 33% and no expression in 14%) (11). These patterns exhibit consistent trends with our results showing a similarly wide range of expression. It is possible that the BTK protein level may reflect strength of BCR signaling in DLBCLs.
In patients with CLL on ibrutinib treatment, a decrease of BTK protein level was observed in CLL cells (26), whereas another report for patients with CLL on acalabrutinib treatment showed variable re-synthesis of BTK affecting target occupancy and BCR reactivation (27). These findings suggest the possible influence BTK-inhibitor treatment to the kinetics of BTK protein. Since the BTK level in neoplastic B-cells may reflect BCR signaling activation, it might affect outcomes of B-cell lymphomas. Our data revealing the validity of BTK positivity as a biomarker tied to IPI in non-GCB DLBCLs remains to be validated further in studies with prospective design. Furthermore, the role of BTK positivity on response to BTK inhibitor also needs to be explored.
Considering the experimental observation that ibrutinib-resistant cell lines of non-GCB DLBCL had higher BCL2 gene expression, combination therapy of ibrutinib and BCL2 inhibitor has been suggested for synergistic effects in killing tumor cells (14, 15). As another resistance mechanism, MYC signaling pathway activation was proposed on the basis of MYC-induced activation of BCR signaling (16). Clinically, BCL2/MYC coexpresser defines a poor prognostic subgroup showing frequent relapse after chemotherapy (28). In our correlation analysis with BCL2/MYC-related markers, BTK intensity correlated with BCL2 intensity and MYC CNG in DLBCLs, suggesting a close functional relationship between BTK and BCL2, while detailed interaction with the MYC pathway needs to be determined.
In survival analysis, BTK positivity and MYC gene translocation exhibited significant association with PFS. Although BTK, BCL2, MYC or BCL2/MYC coexpression was not independent from IPI, the newly defined BTK/BCL2/MYC risk group provided additional prognostic value independently from IPI in considering DLBCL overall and those with the non-GCB subtype, but not in those with GCB DLBCL. Considering the effects of BTK inhibitor on GCB DLBCL cells (29), the clinical significance of BTK in the GCB subtype needs to be evaluated in a larger cohort.
Our BTK/BCL2/MYC-based risk model may partly overlap the ‘double-hit score’ which gives the highest weight to coexpression of MYC and BCL2 (30). Our data showed that 20% of those with BCL2/MYC coexpression belonged to the BTK/BCL2/MYC low-risk group due to lack of BTK positivity (Table V), and this new category predicted clinical outcome with superior statistical significance to BCL2/MYC coexpression, single marker BTK and other BTK-based combined markers. Due to lack of BTK-inhibitor treatment, it is still unclear whether the prognostic effects of BCL2/MYC are mainly related to BCR/BTK signaling. In our cohort, the BTK-positive rate (70.4% for the non-GCB subtype) far exceeds the known response rate (37%) to BTK inhibitor in non-GCB DLBCL (9). Nevertheless, the similar proportions of the complete remission rate (16%) (9) and our BTK+/BCL2−/MYC− subset, i.e., BTK positivity with no activation of potential resistance pathway (14.1%; Table V), might give an insight into further exploration.
In conclusion, BTK positivity correlated with aggressive features including high IPI, advanced stage and shorter PFS, especially in patients with non-GCB DLBCL. The new BTK/BCL2/MYC-based risk model reflecting concurrent activation of BTK and one or more potential resistance-related marker (high-risk) was an independent poor prognostic factor in patients with DLBCL and those with non-GCB subtype. These findings may provide an insight into the pathogenesis of DLBCLs, and contribute to risk stratification in patients with DLBCL.
Acknowledgements
This study was supported by Basic Science Research Program through the National Research Foundation (NRF) of Korea funded by the Ministry of Science and ICT (NRF-2019R1F1A1061920) and Hun Kim family Charitable Foundation through Seoul National University Foundation (800-20190386).
Footnotes
Authors’ Contributions
YBH and JHP designed this study and wrote the article. JMY performed the experiments. YBH and HJK organized clinicopathological data. YBH and JHP performed the statistical analysis. JOL and JSL helped prepare the article. JHP supervised the study and edited the article.
Conflicts of Interest
There are no conflicts of interest regarding this study.
- Received August 20, 2021.
- Revision received September 13, 2021.
- Accepted September 16, 2021.
- Copyright © 2021 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.