The different predictive effects of the intensity and proportion of CD20 expression on the prognosis of B‐lineage acute lymphocyte leukemia

Abstract The prognostic effects of the CD20 positivity have been studied extensively in B‐lineage acute lymphocyte leukemia (B‐ALL) patients, but the results remain controversial. The aim of this study is to investigate the different predictive effects of the intensity and proportion of CD20 expression on the prognosis for B‐ALL patients by retrospective analysis. The mean fluorescence intensity (MFI) and percentage of CD20 on B‐ALL cells from 206 patients with B‐ALL were dynamically measured by flow cytometry, and their optimal cut‐off values were determined using the receiver operating characteristic curve. Changes in MFI and percentage of CD20 at various time points and their relationship with prognosis were analyzed. We found that a low baseline CD20 MFI or high CD20 proportion was significantly associated with shorter 5‐year overall survival and progression‐free survival, and the combination of these two factors could more accurately predict worse survival for B‐ALL patients. Furthermore, low CD20 MFI or a high CD20 proportion had different predictive effects for ALL patients with different clinical characteristics and could serve as an independent risk factor for adverse prognosis. There were significant decreases in both the intensity and proportion of CD20 after recurrence in the absence of rituximab treatment, particularly with CD20 intensity. Notably, the decrease of CD20 intensity after recurrence indicated a more shortened survival time. Finally, we conclude that a low intensity or high proportion of CD20 expression may be used as an indicator for inferior prognosis for B‐ALL patients. CD20 intensity is more likely to be a more universal biomarker for worse prognosis.

disease, the estimated 5-year OS is approximately 10%, and there is no standard treatment. Currently, abnormal karyotype, poor early efficacy including complete remission (CR) in 4 weeks or minimal residual disease (MRD) status within 3 months, high leukocyte counts, and age greater than 35 years are usually considered to be high-risk characteristics in ALL [4,5]. Allogeneic hematopoietic stem-cell transplantation (allo-HSCT) may partially improve the prognosis of these high-risk patients [5]. Therefore, it is necessary to identify more and accurate prognostic biomarkers to screen high-risk patients.
CD20 is a non-glycosylated protein expressed on the surface of normal and malignant B cells, and it is expressed in approximately 40%-50% of precursor lymphocytes [6]. CD20 functions as a calcium channel that influences cell cycle progression and differentiation via downstream signaling pathways, resulting in lower expression of proapoptotic proteins, such as Bax/Bak, and overexpression of anti-apoptotic proteins, such as Bcl-2.
At present, most studies define CD20 positivity (CD20+) as ≥20% of cells expressing CD20 according to NCCN guidelines [4]. CD20+ is often associated with poor prognosis, but this idea remains controversial [7][8][9][10][11]. Recent studies have shown that a CD20 expression rate between 10% and 20% is associated with worse prognosis compared with expression of less than 10% [7,12]. Fortunately, the addition of CD20 monoclonal antibodies such as rituximab to chemotherapy regimens have prolonged the survival of Philadelphia chromosome (Ph)-negative B-lineage ALL (B-ALL) patients who are CD20+ [13]. Additionally, a previous study has shown that fluorescence intensity may better describe the distribution of antigens on leukemia cells than percent positivity [14]. However, little is known about the fluorescence intensity of CD20 expression for prognosis assessments for ALL patients [14].
Thus, this study aims to investigate the predictive effects of the intensity and proportion of CD20 expression on the prognosis of B-ALL patients.

Regimens and responses
ALL patients who were suitable for intensive chemotherapy underwent the CALLG2008 protocol (a protocol developed by the Chinese Acute Lymphoblastic Leukemia Cooperative Group for ALL) [15] including induction, consolidation, maintenance, and CR was defined as BM blasts <5% with no evidence of extramedullary disease and with recovery of peripheral blood counts (absolute neutrophil count >1.0 × 10 9 /L, platelet count >100 × 10 9 /L).
Recurrence was defined as the reappearance of blasts in the blood or BM (≥5%) or any extramedullary site after CR. OS was defined as the time from disease diagnosis to death or last follow-up. Progressionfree survival (PFS) was defined as the time from diagnosis to disease progression or disease-related death.

Calculation of the optimal cut-off values for the proportion and MFI of CD20 expression
The intensity and proportion of CD20 expression on leukemia cells in B-ALL patients were obtained by gating nucleated cells using FAC-SCalibur (Becton Dickinson, San Jose, CA, USA) and by analyzing using Navios Flow Cytometer software (Beckman Coulter, Brea, CA, USA).
And CD20 expression was dynamically monitored at different time points, such as diagnosis, after the first induction therapy, the first recurrence, and multiple relapse. The mean fluorescence intensity (MFI) for CD20 expression was determined by dividing the mean CD20 antibody staining intensity by that of the isotype control. We used OS of B-ALL patients as the discriminator to determine the optimal cutoff value of MFI and proportion of CD20 expression according to the receiver operating characteristic (ROC) curve. The point on the curve where the sum of sensitivity and specificity reached the maximum, namely Youden's index, was the optimal cut-off value for the marker.
According to the ROC curve, the area under the curve (AUC), 95% confidence interval (CI), and optimal cut-off value were determined.

2.4
Statistical methods SPSS

Baseline low MFI and high proportion of CD20 expression had an adverse effect on the prognosis of B-ALL patients
The 206 eligible patients were categorized into high and low groups according to the optimal cut-off values for the CD20 MFI or proportion. The 5-year OS and PFS in the baseline low MFI group were 44.2% and 38.2%, respectively, which were remarkably lower than that in the high MFI group (66.5% and 59.8%; p = 0.016 and 0.042, respectively) ( Figure 2A,B). Similarly, the 5-year OS and PFS of patients with a high baseline CD20 proportion were 48.9% and 40.6%, respectively, which were significantly lower than that for patients with a low proportion (71.2% and 64.2%, respectively; p = 0.015 and 0.007, respectively) ( Figure 2C,D).
In order to further study the superior predictive value of the combination of MFI and proportion on the prognosis of patients with   Figure 2F). These findings suggest that the combination of the CD20 MFI and CD20 proportion may be used to accurately evaluate the prognosis of B-ALL patients ( Figure S1).
In addition, we investigated the effects of the baseline MFI and proportion of CD20 on early efficacy for B-ALL patients. It was found that there were no significant differences in the CR rate in 4 weeks or the MRD negative rate within 3 months between the low and high CD20 MFI groups or the low and high CD20 proportion groups (all pvalues > 0.05). However, CR rates within 4 weeks were significantly higher in Group A than that in Group B (p = 0.042). There was no statistically significant difference in the MRD negative rate within 3 months between the two groups (p > 0.05) ( Figure S2).

Predictive effects of low MFI and high proportion of CD20 expression on the OS of B-ALL patients with various clinical characteristics
The CD20 MFI and proportion demonstrated different predictive effects on OS for patients with different clinical characteristics. As shown in Figure 3A Similarly, as shown in Figure 3B, a higher proportion of CD20 expression was significantly correlated with worse OS for B-ALL patients with these factors including greater than 35 years old, Phnegativity, and no allo-HSCT treatment, which were defined as inferior prognosis (all p-values < 0.05). Furthermore, when patients had <2 gene mutations, baseline WBC <30 × 10 9 /L, PLT <100 × 10 9 /L, or CR in 4 weeks, the higher CD20 proportion might predict shorter OS (all p-values < 0.05). However, the high CD20 proportion was not corre-

3.5
The MFI and proportion of CD20 expression significantly decreased after recurrence and had adverse effects on prognosis To dynamically measure the levels of MFI and proportion of CD20 at diagnosis, after induction, and after recurrence, we found that there was a significant difference in different disease states (p < 0.001 and 0.019, respectively) ( Figure 4A,B). The median CD20 MFI was 23.86 at diagnosis, then decreased to 19.66 after induction, further reduced to 14.06 at the first recurrence, and was 13.05 at multiple recurrences. Contrary to the MFI, the baseline median CD20 proportion was 27.12%, and it mildly increased to 30.30% after induction when the MRD was still available; however, once the disease relapsed, it dropped to 15.68% at the first relapse and 18.83% at multiple relapses.
Moreover, within-group comparisons revealed there was a significant decrease in the CD20 MFI (p < 0.001) and proportion (p = 0.003) after recurrence compared with that before relapse (Figure 4).

F I G U R E 2 Baseline low mean fluorescence intensity (MFI) and a high proportion of CD20 expression had an adverse effect. (A and B)
Low baseline CD20 MFI is remarkably associated with shorter overall survival (OS) and progression-free survival (PFS) (p = 0.016 and 0.042, respectively). (C and D) High baseline CD20 proportion is significantly related to worse OS and PFS (p = 0.015 and 0.007, respectively). (E and F) Low MFI combined with high proportion is significantly associated with shorter OS (p = 0.013) and PFS (p = 0.004) To further explore the impact of decreased CD20 expression after relapse on survival time, 90 patients with recurrent ALL were analyzed and grouped according to the optimal threshold of the MFI and proportion of CD20 expression. The 3-year OS in the low MFI group (8.1%) after recurrence was significantly lower than the high group (32.6%) (p = 0.002) ( Figure 5A). In addition, the 3-year OS of patients with a high CD20 proportion (18.6%) after relapse mildly decreased compared with the 23.3% observed for patients with a low CD20 proportion (p = 0.567) ( Figure 5A).

Low MFI and high proportion of CD20 expression are independent prognostic factors for poor survival
In univariate analysis, low CD20 MFI, high CD20 proportion, ≥35 years old, baseline WBC ≥30 × 10 9 /L, Ph-positive, MRD positive within 3 months, and no CR within 4 weeks were significant risk factors for poor OS and PFS. Moreover, allo-HSCT could significantly prolonged the OS (Table 2).

F I G U R E 3
The predictive effects of the mean fluorescence intensity (MFI) and proportion of CD20 expression on overall survival (OS). (A) Lower CD20 MFI is significantly associated with inferior OS for B-lineage acute lymphocyte leukemia (B-ALL) patients with a hemoglobin (Hb) <90 g/L, platelets (PLT) <100 × 10 9 /L, Philadelphia chromosome (Ph)-negativity, ≥2 gene mutations, complete remission (CR) in 4 weeks, and minimal residual disease (MRD) positivity in 3 months, and regardless gender, age, white blood cells (WBC) count, and allogeneic hematopoietic stem-cell transplantation (allo-HSCT) (all p-values < 0.05). (B) Higher proportion of CD20 expression is significantly correlated with worse OS for B-ALL patients who were greater than 35 years, WBC <30 × 10 9 /L, PLT <100 × 10 9 /L, non-allo-HSCT, Ph-negativity, <2 gene mutations, and CR in 4 weeks (all p-values < 0.05) F I G U R E 4 Changes in the mean fluorescence intensity (MFI) and proportion of CD20 in different disease states. There are significant differences in the MFI (p < 0.001) and proportion (p = 0.019) of CD20 in different disease states. Compared with before recurrence, the MFI (A) (p < 0.001) and proportion (B) (p = 0.003) of CD20 expression significantly decreased after recurrence. The lines represent the median and 95% confidence interval (CI) F I G U R E 5 Correlation between CD20 expression after recurrence and overall survival (OS). (A) OS was significantly lower in the low CD20 mean fluorescence intensity (MFI) group than that in the high MFI group after recurrence (p = 0.002). (B) The OS in the high CD20 proportion group after recurrence mildly decreased compared with the low proportion group (p = 0.567) Statistically significant clinical factors for OS and PFS (p < 0.1) were included in the multivariate analysis, revealing that WBC count ≥30 × 10 9 /L, ≥35 years old, low CD20 MFI, and MRD positive within 3 months were independent risk factors for inferior OS. In contrast, allo-HSCT was a protective factor for long-term survival. Similarly, baseline WBC ≥30 × 10 9 /L, ≥35 years old, high CD20 proportion, no CR in 4 weeks, and MRD positive within 3 months were independent risk factors for worse PFS (Table 3).

DISCUSSION
The prognostic effects of the CD20 expression percentage have been studied extensively in B-ALL patients, but the results remain contro-versial. In this retrospective study, we focused not only on the CD20 proportion but also on the CD20 intensity and found that their different prognostic values, that is, low CD20 MFI or high CD20 proportion, can be used as an independent indicator of inferior prognosis for B-ALL patients, and these were applied to ALL patients with different clinical factors. Moreover, the CD20 MFI combined with the CD20 proportion could more accurately evaluate poor prognosis for B-ALL patients.
There were significant decreases in both the intensity and proportion of CD20 expression after recurrence in the absence of rituximab treatment, particularly for the intensity. Notably, the decrease of CD20 intensity after recurrence indicated a more shortened survival time.
In this study, we first determined the optimal thresholds of the proportion and MFI of CD20. Coincidently, the optimal value for the CD20 proportion was 11.21%, which is similar to the optimal value of 11.7% TA B L E 2 Univariate analysis of risk factors associated with overall survival (OS) and progression-free survival (PFS) found in a previous study in which the high percentage was an independent adverse factor for event-free survival [12]. The effects of a high CD20 percentage on the survival of B-ALL patients has been widely demonstrated [7,12], and the prognostic effects of CD20 intensity have been less studied [14]. To screen for more prognostic biomarkers for B-ALL patients, we first determined that >35 years old, high WBC, no allo-HSCT treatment, no CR in 4 weeks, and MRD positive within 3 months are predictors of inferior outcome, which is consistent with the accepted high-risk index [3,4]. More notably, low CD20 MFI and high CD20 proportion were listed as independent poor prognostic factors for OS and PFS. In addition, lower MFI or higher proportion of CD20 was closely associated with shorter survival for newly diagnosed or refractory/relapsed patients. By contrast, patients with high-intensity CD20 expression hada favorable prognosis.

Variables OS HR (95% CI) p-Value PFS HR (95% CI) p-Value
To specifically elucidate the prognostic value of CD20 expression, we found that the predictive effects of CD20 intensity and CD20 proportion for survival could be applied to ALL patients with different clinical factors. Notably, the prognostic effects of low CD20 MFI can cover a wider range of B-ALL patients than high CD20 proportion. Interestingly, high proportion or low intensity of CD20 were not associated with OS for Ph-positive ALL patients. The underlying reason may be due to the integration of TKIs into chemotherapy regimens [3]. Moreover, the low intensity and high percentage of CD20 have similar predictive effects on worse OS for patients who achieve CR within 4 weeks.
Fortunately, as reported in the literature [16,17], the addition of rituximab to standard chemotherapy can significantly improve the efficacy and prognosis of patients with CD20-positive lymphoid malignancies. Moreover, rituximab-mediated effects depend on CD20 levels with higher intensity expression, leading to higher elimination [17,18].
In addition, some studies have reported that patients with a CD20 expression level below 20% may also benefit from rituximab [7,12].
In this study, we found that patients with low intensity and high percentage of CD20 had the worst prognosis and may be suitable candidates for the use of higher affinity anti-CD20 antibodies, such as obinutuzumab or ofatumumab [6].
Generally, CD20 expression was down-regulated in response to anti-CD20 therapy because of antigenic modulation and gene mutations [19,20] or "shaving" caused by the endocytosis of monocytes and macrophages [21,22]. However, this study found there is a significant decrease both in the intensity and proportion of CD20 expression after recurrence in the absence of rituximab treatment, and the underlying reasons remain unclear. Recent studies have reported that patients with NOTCH1 mutations are chemorefractory and associated with lower CD20 MFI in B-cell tumors [23][24][25]. Moreover, expression of the MS4A1 gene encoding CD20 was negatively regulated by the transcription factors FOXO1, CREM, EZH2, and MYC, and the decrease in CD20 expression was commonly found in relapsed/refractory non-Hodgkin lymphomas [6,[26][27][28][29]. Furthermore, the decrease in CD20 expression after recurrence may be due to the selective pressure caused by the expansion of chemoresistant subclones [30]. Nevertheless, it appears that decreased CD20 expression may be associated with the worst prognosis and not just CD20 antibody privilege, which may be related to chemoresistance [31].
This study is an extension and supplement for traditional biomarkers. Some of the limitations of this study include its retrospective design and the absence of control studies including treatment with rituximab. Further prospective studies are needed in the future.
Our findings suggest that not only a high CD20 proportion but also a low CD20 intensity are significantly associated with worse OS and PFS for B-ALL patients. The decrease of CD20 intensity after recurrence indicated a more shortened survival time. It should be noted that CD20 intensity is more likely to be a more universal biomarker for adverse prognosis than CD20 proportion.

This study was supported by the Foundation for Young Teachers' Basal
Research of Zhengzhou University (jc202050015).

FUNDING INFORMATION
Foundation for Young Teachers' Basal Research of Zhengzhou University (jc202050015).

CONFLICT OF INTEREST
The authors declare they have no conflicts of interest.

ETHICS STATEMENT
The ethics committees of Affiliated Cancer Hospital and Affiliated People's Hospital of Zhengzhou University have approved this study.

PATIENT CONSENT STATEMENT
Informed consent was obtained from all participants included in the study.

AUTHOR CONTRIBUTIONS
Yun Tian collected and analyzed the data, and wrote the paper. Xiaojiao Wang, Hao Ai, Xiaodong Lyu, and Qian Wang collected samples and provided clinical data. Xudong Wei and Yongping Song reviewed the final version of the manuscript and assisted in the critical review of the manuscript and data. Qingsong Yin designed the research study, wrote the paper, and provided critical revisions. All authors reviewed the manuscript.