Re-Evaluate of DNA Index Among the Childhood Acute Lymphoblastic Leukemia

Background This study was aimed to re-evaluate the value of DNA index(DI) among pediatric acute lymphoblastic leukemia (ALL) with the recent chemotherapy protocols. Methods Retrospective study were analysis among pediatric ALL patients from the TAGET dataset. Result Totally, 1,668 eligible pediatric patients were enrolled in this study. Of them, 993 are male and 675 are female with a median age of 7.6 years old. The median follow-up for those patients was 7.7 years (range 0.1-15.7 years). The probability of 10-year EFS and OS were reported to be 68.3±2% and 79.6±2%, respectively. Multivariate analysis showed that DI was signicantly associated with better EFS and OS. The threshold effect of DI on poor outcome was signicant after adjusting for potential confounders. The adjusted regression coecient (Log RR) was 0.7 (95%CI 0.1-3.2, P=0.597) for DI <1.1 while 8.8 (95%CI 1.4-56.0, P=0.021) for DI ≥ 1.2 and 0.0 (95%CI 0.0-0.8, P=0.041) for 1.1 ≤ DI<1.2. Generalized additive models revealed that the lowest rates of the adverse outcomes estimated to occur among DI between 1.1 and 1.2. Conclusion For those childhood ALL treated with recent chemotherapy protocols, the DI between 1.1 and 1.2 can serve as a signicant cut-point discriminating the risk group, which indicated an independent prognostic factor.


Background
Acute lymphoblastic leukemia (ALL) is the most common hematological malignant tumor in children, accounting for 80% of childhood leukemia [1] . ALL is a malignant clonal disease of hematopoietic stem cells, which is a heterogeneous hematological disease caused by abnormal clone and proliferation of primitive and immature lymphocytes in bone marrow and peripheral blood. With the improvement of the understanding of molecular genetics and pathogenesis of ALL, combined with risk classi cation, multidrug intensive therapy and hematopoietic stem cell transplantation, the cure rate and survival outcome of children with leukemia have been signi cantly improved in recent years, but the pathogenesis and speci c etiology are still unclear [2][3] . So far, it has been found that cytogenetic changes play an important role in the occurrence and development, treatment and prognosis of childhood leukemia [4] .Among these cytogentic ndings, the DNA index (DI), which represents the DNA content of leukemic cells, has been considered an important prognostic factor for risk determination [5] . However, studies [6][7][8][9] on the best DI threshold affecting prognosis were published in 1980s, 1990s and 2008, while the latest study [10] in 2017 suggested the value of DI and its conventional cut-point should be reevaluated in patients treated with the recent chemotherapy protocols. In this study, our purpose was to reevaluate the value of DI on early treatment response and prognosis among 1,668 pediatric ALL patients from the therapeutically applicable research to generate effective treatment (TARGET) dataset with the recent chemotherapy protocols, so as to provide evidence for assessment of risk strati cation and differences in the intensities of chemotherapy in treatment protocol. Page

Study Participants
The clinical data about pediatric ALL aged≤18 years old was downloaded from the TARGET dataset (December 21, 2020) (https://ocg.cancer.gov/ programs/target/data-matrix). Between May 2000 and August 2015, 1,990 patients were registered in the TARGET database: 50 were not eligible (second malignancy, n=17; Down syndrome, n=33), 272 were not assessable (inadequate information on diagnosis or treatment), and 1,668 childhood ALL were enrolled in our study nally. The trial was approved by the Ethics Committee, Sun Yat-sen Memorial Hospital, Sun Yat-sen Memorial University, and informed written consent was obtained from each patient's guardian.Informed consent was obtained in accordance with the Declaration of Helsinki. All methods were performed in accordance with the relevant guidelines and regulations such as French-American-British (FAB) criteria in the manuscript.

Diagnostic Studies
Bone marrow(BM) samples were transported overnight and processed at the AIEOP reference laboratory within 24 hours. Diagnosis of ALL was based on French-American-British (FAB) criteria: less than 3% blasts positive for myeloperoxidase or Sudan black, and negative for nonspeci c esterase [11] .
Immunophenotyping was performed by ow cytometry (Epics XL; Beckman Coulter, Miami, FL) [12][13] . For DI measurement, 300,000 to 500,000 monocytes were stained with propidium iodide using automatic DNA staining equipment and analyzed by ow cytometry after two hours. DNA histograms obtained using semiautomatic cell cycle analysis software (Multicycle; Phoenix, San Diego, CA) allowed compensation for doublets and overlapping nuclei. DI was de ned as the ratio of the number of mode channels of G0/G1 peak in tumor cells to that in normal cells. The donor lymphocytes served as the normal control, and the residual normal T lymphocytes in the bone marrow samples served as the internal control [14] .
Complete remission (CR) was de ned as non-leukemic signs, no leukemic cells detected in blood smears, active hematopoiesis in bone marrow, less than 5% of leukemic primordial cells, and normal cerebrospinal uid (CSF). Bone marrow aspiration was examined on Day+29 at the end of phase IA. According to the dataset, the year of diagnosis is from 2000 to 2015. The last follow-up year was from 2000 to 2017. All cases were diagnosed and subtypes were classi ed according to FAB classi cation.
The chemotherapy regimen of ALL includes 9906, AALL0232, AALL0331 and AALL0434, all of which include the remission induction phase followed by an intensi cation phase. All study participants received written informed consent. Informed consent was obtained in accordance with the Helsinki Declaration. Cytogenetic in situ analysis was performed by standard G-banding / uorescence in situ hybridization (FISH) techniques.

Statistical analysis
Baseline characteristics were presented as mean±SD for continuous variables and as frequency (%) for categorical variables. Comparisons between groups were made using the chi-square test for categorical variables and analysis of variance or the Kruskal-Wallis test for continuous variables. To assess outcome, the following parameters were used: CR rate, event-free survival (EFS, de ned as time between diagnosis and rst event, including induction failure, relapse, progression or death of any cause), overall survival (OS, de ned as time between diagnosis and death from any cause). EFS and OS were estimated by the Kaplan-Meier method and compared using the log-rank test. The univariate and multivariate logistic regression analyses were performed to identify risk factors of early treatment response and prognostic factors. A generalized additive model (GAM) with a spline smoothing function was applied to examine the relationship between DI and adverse clinical outcomes (adverse clinical outcomes, de ned as time between diagnosis and rst event, including induction failure, relapse, progression or death of any cause).
Adjusted OR with 95% CIs were estimated to evaluate the association of DI and adverse clinical outcomes, which was with adjustments for gender, age, race, chemotherapy protocol, white blood cell(WBC); central nervous system(CNS) status, fusion gene, prednisone response, BM blasts on Day+29, immunophenotype and karyotype. All statistical analyses were performed using the IBM SPSS Statistics version 22.0, and EmpowerStats (http://www.empowerstats.cn/). A 2-tailed p<0.05 was considered to be statistically signi cant in all analyses.

Baseline characteristics of patients
Totally, 1,668 eligible pediatric patients were enrolled in this study. The baseline characteristics of the eligible patients were presented in Table 1. Of them, 993 (59.5%) are male and 675 (40.5%) are female with a median age of 7.6 years old. According to the immunophenotype, 721 (43.2%) patients were diagnosed as B-precursor ALL.In addtion, Fusion gene was detected in 1,307 patients via conventional Gbanding analysis and the most common positive fusion gene was ETV6/RUNX1. Among the pediatric patients, the median of initial WBC was 33.7×10 9 /L (range 0.4-1306.0×10 9 /L), and the median DI was 1.0 (range 0.0-1.9).
As a result, 222 patients accept 9906 protocol, 789 patients accept AALL0232, 415 patients received AALL0331 protocol and the remaining 242 patients received AALL0434 protocol. As for the early treatment response, 1,154 patients responded well to prednisone on Day+8, while the remaining 514 patients did not respond well to prednisone. Most of the pediatric patients (n=1,625) achieved CR at the end of remission induction chemotherapy (Day +29).
The median follow-up for those patients was 7.7 years (range 0.1-15.7 years). Finally, 309 patients subsequently developed hematological relapse (presence of leukemic blasts > 5% in BM), 113 patients developed CNS relapse and testes site of relapse occurred in 11 children.The probability of 10-year EFS and OS were reported to be 68.3 ± 2% and 79.6 ± 2%, respectively (Fig 1).

Multivariate analysis of prognostic factors
When we included DI with other risk factors in the Cox model, including age, gender, race, WBC, fusion gene, prednisone response, BM blasts Day+29, immunophenotype and karyotype as co-variables, we identi ed DI as an independent factor for both EFS and OS in pediatric patients with ALL (Table 3). DI was signi cantly associated with better EFS (HR=0. 3.4 The value of DI cut-point among prognostic impact of pediatric ALL Generalized additive models (Fig 2) was used to visually assess functional relationships between DI and the risk of adverse clinical outcomes. This analysis was conducted using both logarithmic transformed and untransformed data. Log (relative risk) can be converted to a relative risk by taking antilog. For example, a log (relative risk) of 0 implies the relative risk of 1 (no impact on the probability of having adverse clinical outcomes), whereas a log (relative risk) of 1 implies the relative risk of 2.71 (ie, 2.71-fold increase in the probability of having adverse clinical outcomes). After adjusting for these possible factors related to adverse clinical outcomes, including gender, age, race, chemotherapy protocol, WBC, CNS status, fusion gene, prednisone response, BM blasts day+29, immunophenotype and karyotype, the Ushaped relationships between DI and adverse clinical outcomes were con rmed in multivariable analyses. The threshold effect of DI on poor outcome was signi cant after adjusting for potential confounders. The adjusted regression coe cient (Log RR) was 0.7 (95%CI 0.1-3.2, P=0.597) for DI <1.  (Table 4). Analyses showed that the lowest rates of the adverse outcomes estimated to occur among DI between 1.1 and 1.2. Moreover, a poor outcome signi cantly increased with increasing DI after the turning point (DI≥1.2) (Fig 2).

Survival analysis of DI cut-point
Base on DI cut-point of 1.1 and 1.2 obtained above, we analyzed the survival as just three groups having different DI ranges ( DI<1.1, 1.1≤DI<1.2 and DI≥1.2). As a result, the EFS of pediatric ALL with a DI between 1.1-1.2 were higher than those with DI of <1.1 or ≥1.2 (10-year EFS, 72.6±6% versus 67.7±2%, P=0.15), but no signi cant difference was found. However, the OS of pediatric ALL with a DI between 1.1-1.2 were signi cantly higher than those with DI of <1.1 or ≥1.2 (10-year OS, 88.9±4% versus 78.3±3%, P<0.05) (Fig 3, Fig 4).

Discussion
The TARGET ALL project team consists of several Children's Oncology Group(COG) investigators from different institutions. After the inclusion and exclusion criteria, 1,668 pediatric ALL were eventually enrolled in our study, which was the largest-scale study to re-evaluate the value of the DI among childhood ALL treated with recent chemotherapy protocols since 2000s.
First of all, we investigated the value of DI in early treatment response of childhood ALL. In the existing evaluation system, the main factors affecting the early treatment response of childhood ALL include age, initial WBC, speci c gene fusions and so on. According our univariate and multivariate analysis, which showed that age≥10 years, BCR/ABL1 fusion gene and poor prednisone response had a signi cant negative impact on early treatment response, while hyperdiploid and an improved chemotherapy regimen had a signi cant positive impact, and these results were basically consistent with those reported in previous literature [15][16][17][18][19] . It has been reported [20] that a low DNA index was bene cial for early treatment response. However, no association was found between DI and early treatment response.Based on the above results, we implied those low DI population might both had higher WBC and older age that lead to the correlation was not obvious.
ETV6/RUNX1 positive ALL is considered to occur in prenatal period, which may precede the preleukemic stage [21] . Furthermore, the existence of this fusion transcript changes the differentiation process of hematopoietic progenitor cells and enhances the self-renewal of hematopoietic progenitor cells, especially B-line hematopoietic progenitor cells [22] . Wang [23] et al reported that the induced remission rate of 77 ETV6/RUNX1 positive B-ALL children was 100%, and the 5-year EFS and OS were 90% ±3% and 96% ±3%, respectively. Based on the favorable molecular response to treatment and good clinical outcomes, this rearrangement is considered to be of signi cant therapeutic signi cance. Results of multivariate analysis from our study showed ETV6/RUNX1 fusion gene signi cantly associated with better EFS and OS. These ndings were consistent with a pediatric study of previous studies [23][24][25] .
Moreover, we showed an independent favorable outcome for the patients on DI in terms of 10-year EFS and OS.We further revealed a threshold effect based on the DI and provide clear evidence of a nonlinear association between DI and adverse clinical outcomes. To the best of our knowledge, this is the rst study to describe a U-shaped relationship between DI and adverse clinical outcomes. Interestingly, using these data, we were able to identify DI ranging from 1.1 to 1.2 in which the rates of an adverse clinical outcome of induction failure, relapse, progression or death of any cause were lowest in our population. Next, we explored the survival as just three groups having different DI levels among pediatric patients with ALL. Finally we found that DI between 1.1 to 1.2 conferred favorable prognostic impact on survival in patients with DI of 1.1 or ≥1.2. Inconsistent with previous studies [6][7][8][9] , DI of ≥1. 16 were not a signi cantly associated with high EFS and OS in our study but DI of 1.1-1.2 did. In addition, Noh [10] et al reported that the group with a DI of 1.00-1.90 showed signi cantly higher OS and EFS than the group with a DI of <1.00 or >1.90, whereas the DI of 1.16 was not a signi cant cut-point in discriminating the risk group among children.We speculated that the reason for the inconsistency with our results may be that the number of cases included in Noh's studies was too small and was limited to pre-B. Secondly, our pediatric patients were treated with the recent chemotherapy protocol. Furthermore, we used smooth curve tting after adjusting the confounding factors, which greatly increases the reliability of our results. Taken together, our study revealed that the DI of 1.1-1.2 was a signi cant cut-point to evaluate the prognostic value among childhood ALL, although pediatric patients were treated with recent treatment protocols. This information will be useful in the application of evidence that those DI out of 1.1-1.2 should targeted adjustment of chemotherapy program, reduce complications, and pay attention to close follow-up.
The study has several limitations. Most importantly, it was a retrospective study, so bias in this study was inevitably. Secondly, the lack of some data may lead to the incompleteness of the results. In spite of these limitations, this study is a multicenter, large-scale retrospective study to re-evaluate the value of DI, which provide strong evidence for those pediatric ALL treated with recent chemotherapy protocols. In conclusion, the DI between 1.1 and 1.2 can serve as a signi cant cut-point discriminating the risk group as ever, which remained an independent prognostic factor.

Declarations Ethics approval and consent to participate
The research protocol and informed consent were approved by the ethics committee of Sun Yat-sen Memorial Hospital of Sun Yat-sen University and cooperation agreements were signed with all collaborative members. Consent to participate was obtained from the participants and the parents/guardians of the children under the age of 18.

Consent for publication
Not applicable.

Availability of data and materials
The data sets used and/or analysed during the current study are available in TARGET dataset (https://ocg.cancer.gov/ programs/target/data-matrix).

Competing interests
The authors declare that they have no competing interests.

Funding
This work was supported by the Guangzhou Science and Technology Program key projects (No.201803010032).