Coexpression of CD5 and CD43 predicts worse prognosis in diffuse large B‐cell lymphoma

Abstract Both CD5 and CD43 are expressed on the surface of B lymphocytes of definite phase and associated with the adverse outcome in diffuse large B‐cell lymphoma (DLBCL). However, the relationship between CD5 and CD43 expression and the prognostic value of CD5/CD43 coexpression in DLBCL are unknown. We herein determined the correlation between CD5 and CD43 expression, as separate factors or in combination, with the clinicopathological features and survival of 200 patients with DLBCL receiving standard chemotherapy with or without rituximab. Among these DLBCL patients, CD5 expression, CD43 expression, and CD5/CD43 coexpression were detected in 18 (9%), 57 (27%), and 10 (5%) patients, respectively, and all were positively correlated with advanced age and nongerminal cell type. CD5‐positive and CD43‐positive DLBCL patients had poorer event‐free survival (EFS, P < 0.001) and overall survival (OS, P < 0.001) than CD5‐negative and CD43‐negative patients, respectively. CD5/CD43 coexpression was correlated with a significantly worse prognosis than CD5 or CD43 expression alone. Univariate analysis showed that CD5 expression, CD43 expression, and CD5/CD43 coexpression were all adverse prognostic factors for DLBCL patient survival, and CD5/CD43 coexpression was associated with a greater relative risk for recurrence and death than either CD5 or CD43 expression alone. Multivariate analysis demonstrated that CD5/CD43 coexpression was an independent prognostic factor for EFS (P < 0.001) and OS (P < 0.001) in DLBCL. In conclusion, our data indicate that DLBCL patients with CD5/CD43 coexpression represent a specific subgroup with a significantly worse prognosis than those expressing either marker alone.

of identifying biomarkers to predict the clinical outcome of high-risk groups or a specific subtype of DLBCL.
The recommended prognostic factors for DLBCL by the International Prognostic Index (IPI) include age >60 years, elevated serum lactate dehydrogenase (LDH), Eastern Cooperative Oncology Group (ECOG) performance status ≥2, stage III or IV, and number of involved extranodal sites >1. However, these five risk factors are not related to the biological features. 5 According to the results of gene expression profiling studies, DLBCL can be stratified into germinal center B-cell (GCB)-like and activated B-cell (ABC)-like or non-GCB-like subtypes, and DLBCL patients with the ABC subtype have an inferior prognosis, 6 which is improved by the addition of rituximab to anthracycline-based regimens. Moreover, increased expression of BCL2 family members plays a role in the resistance of DLBCL to chemotherapy. 7,8 BCL6 was reported to be associated with a better prognosis, and patients with BCL6-positive DLBCL experienced relatively favorable outcomes after treatment with the CHOP regimen. 9 Although these biomarkers may predict the prognosis of DLBCL, few of them have been translated into clinical practice. 10 CD5 is a cell surface glycoprotein and is typically expressed on T cells and a subset of normal naïve B cells as well as lymphoma cells, mainly chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL) and mantle cell lymphoma. [11][12][13] Anatomical localization, gene usage, and function are different between CD5-positive B cells and CD5-negative conventional B cells. 14-16 CD5-positive B cells synthesize immunoglobulin (Ig) M autoantibodies, and increased numbers of CD5-positive B cells are associated with some types of autoimmune disease. [17][18][19] CD5 is also expressed in 5%-10% of de novo DLBCL cases, [20][21][22] and CD5+ DLBCL has been included as an immunohistochemical subgroup in the fourth edition of the World Health Organization (WHO) classification. 12 Several studies have demonstrated that DLBCL patients with CD5-positive expression have a poorer overall survival (OS) rate than those without CD5 expression, regardless of the use of rituximab. 23,24 CD43 is a multifunctional type I transmembrane protein that regulates multiple cellular functions, such as cell signal transduction, cell adhesion, activation, proliferation, cell survival, and apoptosis. [25][26][27] CD43 is expressed on the surface of most hematopoietic cells, some lymphomas, and leukemias. 28 CD43 is also expressed in a variety of solid tumors, but is undetectable in normal tissues and benign lesions. 29,30 In particular, the expression level of CD43 glycoforms in cancer cells correlates with the progression stage of the disease. 30 It has been suggested that coexpression of CD43 and CD20 on peripheral B cells is associated with malignancy. 31 Several studies have showed that CD43 is expressed in approximately 25% of DLBCL cases and is an independent adverse prognostic marker for DLBCL. [32][33][34] Both CD5 and CD43 have been found to be expressed on the surface of B lymphocytes of definite phase and associated with clinical outcomes of DLBCL patients. It was suggested that combined detection of CD43 and CD5 could differentiate cancer cells from normal T and B cells. 35 However, to our knowledge, there has been no report about the prognostic value of CD5 and CD43 coexpression in DLBCL. In this study, we analyzed the association between CD5 and CD43 expression alone or in combination with the clinicopathological features and survival of patients with DLBCL.

| Ethics statement
This study was approved by the Research Ethics Committee of the First Hospital of Jilin University (Changchun, China). All patients provided written informed consent.

| Patient cohort
We enrolled 200 Chinese patients with DLBCL diagnosed at the First Hospital of Jilin University from 1 January 2004 to 30 December 2015. The medical records of the patients were retrieved and reviewed. Diagnosis of DLBCL was verified by two professional hematopathologists according to the World Health Organization (WHO) classification system. 2 Patients who had a history of low-grade B-cell lymphoma, human immunodeficiency virus or human T-cell lymphotropic virus type I infection, primary mediastinal, cutaneous B-cell lymphoma, or central nervous system were excluded from this study. All patients received either CHOP or R-CHOP (CHOP plus rituximab) for a median of 5 cycles (3-8 cycles) and with a median follow-up time of 62 months (range, 15-137 months).

| Statistical analysis
All data were analyzed using SPSS 18.0 software (SPSS Inc., Chicago, IL, USA). Associations between CD5 expression, CD43 expression, CD5/CD43 coexpression, and clinicopathological characteristics were determined using the chisquare test. Correlation between CD5 and CD43 expression was also evaluated by the chi-square test. EFS was calculated from the date of diagnosis to the date of documented disease progression, relapse, death, or study termination. OS was calculated from the date of diagnosis until death from any cause or the last follow-up. EFS and OS were estimated using the Kaplan-Meier method and compared by the log-rank test. Univariate analyses, multivariate analyses, and stratified analysis were performed using the Cox proportional hazards regression model. P ≤ 0.05 was set as indicative of a significant difference.

| Patient characteristics
The characteristics of the patients including 109 men and 91 women with a median age of 58.0 years (range, 7-88 years) are summarized in Table 1. Sixty-seven patients (33.5%) were of the GCB subtype, and 133 patients (66.5%) were of the non-GCB subtype with 143 (71.5%) with IPI scores of 0-2 and 57 (28.5%) with IPI scores of 3-5. The patients in clinical stages I-II and stages III-IV were 116 (58%) and 84 (42%), respectively. One hundred and eighteen (59%) were treated with the CHOP regimen, and 82 patients (41%) were treated with R-CHOP regimen.

| Expression of CD5 and CD43 and the correlation between CD5 and CD43 expression in DLBCL
Expression of CD5 and CD43 in DLBCL tissues was assessed by immunohistochemical staining. Positive staining for CD5 and CD43 expression was detected in 18 (9%) and 54 (27%) cases, respectively; CD5/CD43 coexpression was found in 10 (5%) cases (Table 2). Figure   ( Figure 1C), and CD5+/CD43+ ( Figure 1D) in the DLBCL tissues. The expression levels of CD5 and CD43 were significantly correlated due to conegativity (P = 0.004, chi-square test; Table 2). Most expression of CD5 and CD43 was found in overlapping cell populations, and the CD5+/CD43+ group showed a relatively higher Ki-67 index than the other groups ( Figure 1D). Among the 200 DLBCL patients, 8 were positive for CD5 only, 44 were positive for CD43 only, 10 were positive for both CD5 and CD43, and 138 were negative for either CD5 or CD43 expression.

| CD5 expression, CD43 expression, and CD5/CD43 coexpression are prognostic factors of DLBCL independently of chemotherapy with or without rituximab
To assess whether rituximab treatment affected on the prognostic impact of CD5 expression, CD43 expression, and CD5/ CD43 coexpression in DLBCL, the patients were divided into CHOP and R-CHOP treatment groups. The results showed that CD5 expression, CD43 expression, and CD5/CD43 coexpression were all associated with shorter OS (P < 0.001) in both the CHOP and R-CHOP groups (Figure 3).

| Univariate and multivariate survival analyses
To determine the prognostic value of the clinicopathological factors in combination with CD5 and CD43 expression, Expression (C, F) in DLBCL. Survival was significantly better for patients negative for CD5 (P < 0.001 for both EFS and OS) and for CD43 (P < 0.001 for both EFS and OS) than that for patients with positive expression level. Patients with the CD5+/CD43+ coexpression profile had the worst outcome for OS among the 4 groups. Pairwise comparisons showed that a statistically significant difference in survival rates existed between the CD5+/CD43+ group and any of the other three groups (P < 0.05 for both EFS and OS) univariate Cox regression models were applied. In the univariate survival analysis, advanced age (>60 years), advanced Ann Arbor stage, more extranodal involvement (≥2), elevated LDH, high Eastern Cooperative Oncology Group (ECOG) performance status (PS; ≥2), high IPI, non-GCB phenotype, high Ki-67 index (≥80%), CD5 expression, CD43 expression, and CD5/CD43 coexpression were all significant prognostic factors for poor EFS and OS. The relative risk (RR) was estimated by Cox regression. CD5/CD43 coexpression increased the RR for both recurrence and death compared with CD5 or CD43 expression alone (Table 4).
Clinicopathological factors at the 0.10 level in the univariate analysis, including age, Ann Arbor stage, extranodal involvement, LDH, ECOG PS, non-GCB phenotype, and CD5/ CD43 coexpression, were entered into a multivariate survival analysis model. Treatment was also entered into the model for CHOP and R-CHOP, which have long been known to be associated with differences in survival. The results demonstrated that advanced age, advanced Ann Arbor stage, evaluated LDH level, treatment without rituximab, and CD5/CD43 coexpression were independent unfavorable prognostic factors for both EFS and OS ( Table 5).
The associations between CD5/CD43 coexpression and DLBCL survival were further evaluated by a stratified analysis of age, Ann Arbor stage, LDH level, and treatment. As shown in Table 6, the prognostic value of CD5/CD43 coexpression was independent of any of these factors. Moreover, the adverse effect of CD5/CD43 coexpression was more prominent in patients with a normal LDH level (adjusted RR = 42.557 for EFS, adjusted RR = 10.736 for OS).

| DISCUSSION
The present study demonstrated that the expression rates of CD5 and CD43 in DLBCL were 9% and 27%, respectively. Either CD5 or CD43 expression was correlated with advanced age (>60 years), evaluated LDH, B symptoms, non-GC phenotype, and DLBCL mortality. Moreover, in comparison with those for CD5− and CD43− patients, the OS and EFS rates for CD5+ and CD43+ patients were significantly lower. Our data further support the previous observation that either CD5 or CD43 expression can predict poor prognosis in DLBCL. 20  The simultaneous expression of CD5 and CD43 on the surface of B lymphocytes of definite phase has led to the suggestion that combined detection of CD43 and CD5 might differentiate cancer cells from normal T and B cells. 35 To test whether there is any correlation between CD5 and CD43 expression in DLBCL, we performed chi-square test and found that the expression of CD5 and CD43 was strongly associated with most of them expressed in overlapping cell populations. Further analysis of the correlation of CD5/CD43 coexpression with the clinicopathological characteristics of DLBCL showed that CD5/CD43 coexpression was significantly associated with advanced age, gender (male), more extranodal involvement, high IPI, high Ki-67 index, non-GC phenotype, and DLBCL mortality. Survival analysis showed that the patients with CD5/CD43 coexpression had a significantly worse prognosis than either single-positive group or bothnegative group. CD5/CD43 coexpression increased the RR for both recurrence and death compared with CD5 or CD43 expression alone. Moreover, we found that CD5 expression, CD43 expression, and CD5/CD43 coexpression were all significantly associated with the non-GC phenotype, and the patients with CD5/CD43 coexpression were all non-GC phenotype. Our results suggest that CD5/CD43 coexpression cases may represent a specific subset of DLBCL with even more inferior prognosis. CD5+CD43+ DLBCL may be derived from CD5+CD43+ B cells during B-cell development or the malignant transformation of normal B cells. B cells are divided into conventional B (B-2) and B-1 cells, with B-1 cells predominantly produced during fetal and neonatal development. [39][40][41] In mice, B-1 cells are further classified into B-1a and B-1b cells with B-1a cells expressing CD5. CD5 is also expressed on regulatory B cells that involve in immune evasion. 42 [48][49][50][51] These B-1a-like cells may be the origin of the CD5+CD43+ cells in DLBCL, which warrants further investigation. CD5 and CD43 may be involved in the pathogenesis of DLBCL through a variety of mechanisms. CD5 can promote B-cell survival through autocrine production of interleukin (IL)-10. 52 In addition, CD5 maintains B-cell survival by modifying intracellular calcium mobilization and modulating various signaling pathways including ERK1/2, PI3K, and calcineurin. 53 It was also reported that CD5+ DLBCL cases often have complex chromosomal aberrations, 54 suggesting a potential role for CD5 in the regulation of chromosome stability in B cells. CD43 has been shown to play an important role in cell cycle progression, apoptosis, and immune response of B lymphocytes. 55,56 CD43 also was reported to modulate cellto-cell adhesion between hematopoietic cells. 27 Moreover, CD43 can abrogate contact inhibition of cell growth through a molecular mechanism that involves AKT-dependent Merlin phosphorylation and degradation. 57 A mouse study suggested that anti-UN1/CD43 antibody has antitumor activity in UN1positive HPB-ALL lymphoblastoid T cells. 58 Consistently, a recent study showed that patient-derived antibody recognizes a unique CD43 epitope expressed on all AML and has antileukemia activity in mice. 59 However, the mechanism of the coexpression CD5 and CD43 in the initiation, progression, and maintenance of DLBCL is currently unknown and needs further investigation.
Although our results demonstrate the significance of CD5/CD43 coexpression on multivariate analysis, there are some weaknesses in our study. In particular, there were only 10 cases with CD5/CD43 coexpression in our study, and thus, even smaller case numbers were available when patients were separated based on the CHOP or R-CHOP regimen. This low case number may increase the bias in our study. Therefore, the prognostic significance observed in the present study needs to be validated by future independent studies. In summary, DLBCL with CD5/CD43 coexpression may be a clinicopathological variant of DLBCL. However, the origin of CD5/CD43-coexpressing B cells and the precise mechanisms by which coexpression of CD5/CD43 alters the behavior of DLBCL are unclear. Given that the majority of DLBCL patients with CD5/CD43 coexpression are older with a poor performance status and that the addition of rituximab to routine chemotherapy does not improve the outcomes of these patients, it will be important to better characterize this subset of DLBCL. T a b l e 6 . Stratified analysis for CD5/ CD43 coexpression and DLBCL patients' survival