Low‐Dose Cytarabine Plus G‐CSF Induction in Pediatric M2 Acute Myeloid Leukemia: A Multicenter Study

ABSTRACT

This retrospective secondary analysis of the multicenter randomized CALS III‐AML18 trial evaluated the efficacy and safety of low‐dose chemotherapy (LDC) compared with standard‐dose chemotherapy (SDC) in children with M2 acute myeloid leukemia (AML‐M2). A total of 221 children were included, with 109 in the LDC arm and 112 in the SDC arm. The complete remission (CR)/CR with incomplete recovery (CRi) rates after induction I and II were 67.0% versus 70.5% (p = 0.568) and 84.4% versus 89.3% (p = 0.740), respectively. The 3 year overall survival (OS) rates were 81.2% ± 3.9% and 86.9% ± 3.3% in the LDC and SDC arms, respectively (p = 0.304). The corresponding 3 year relapse‐free survival (RFS) rates were 82.0% ± 3.8% and 88.1% ± 3.1% (p = 0.205), and the 3 year event‐free survival (EFS) rates were 62.1% ± 4.8% and 70.8% ± 4.4% (p = 0.218). Among patients with KIT mutations, outcomes were poorer in the LDC arm than in the SDC arm, including 3 year OS (67.1% ± 8.7% vs. 91.5% ± 4.1%, p = 0.011), RFS (70.6% ± 7.8% vs. 87.5% ± 4.8%, p = 0.038), and EFS (49.2% ± 8.7% vs. 81.0% ± 5.7%, p = 0.002). Among patients who developed sepsis, grade 3–5 events were less frequent in the LDC arm than in the SDC arm (33.3% vs. 78.8%, p < 0.001). LDC was also associated with faster neutrophil and platelet recovery, lower transfusion requirements, and reduced treatment costs during induction. No statistically significant differences in remission or survival outcomes were observed between LDC and SDC, while the LDC regimen showed a lower treatment‐related burden. These findings support further evaluation of LDC as a toxicity‐sparing induction strategy in selected pediatric patients with AML‐M2. The poorer outcomes observed among patients with KIT mutations warrant caution when considering treatment de‐intensification in this subgroup.

Abbreviations

AML

acute myeloid leukemia

AML‐M2

M2 subtype of AML

Ara‐C

cytarabine

CI

confidence interval

CR

complete remission

CRi

CR with incomplete recovery

EFS

event‐free survival

G‐CSF

granulocyte colony‐stimulating factor

HSCT

hematopoietic stem‐cell transplantation

LDC

low‐dose chemotherapy

MRD

measurable residual disease

NR

no remission

OS

overall survival

PR

partial remission

RFS

relapse‐free survival

RR

risk ratio

SDC

standard‐dose chemotherapy

1. Introduction

Acute myeloid leukemia (AML) accounts for approximately 15%–20% of childhood leukemia. With advances in risk‐adapted therapy and supportive care, survival rates have approached 80% in contemporary pediatric AML, and complete remission (CR) rates exceed 90% in many treatment protocols [1, 2]. Nevertheless, intensive chemotherapy incorporating cytarabine (Ara‐C) and anthracyclines remains the backbone of treatment for most children with AML and is associated with substantial myelosuppression and infectious complications [3, 4]. Thus, reducing induction‐related toxicity without compromising antileukemic efficacy remains an important therapeutic goal.

Low‐dose Ara‐C combined with granulocyte colony‐stimulating factor (G‐CSF) has been explored as a less myelosuppressive induction strategy [5]. Pediatric studies have suggested that reduced‐dose induction can retain antileukemic activity, and the recently reported multicenter randomized CALS III‐AML18 trial demonstrated noninferior remission rates and similar long‐term survival with low‐dose induction compared with standard‐dose chemotherapy, together with faster neutrophil and platelet recovery [6, 7]. However, AML is biologically heterogeneous, and these overall findings do not establish whether treatment de‐intensification is equally appropriate for specific AML subtypes [8]. The M2 subtype of AML (AML‐M2) is frequently associated with t(8;21) (q22;q22), resulting in the AML1‐ETO fusion, and relatively high G‐CSF receptor expression has been reported in M2 blasts, providing a biological rationale for G‐CSF–primed low‐dose Ara‐C in this subgroup [9]. Nevertheless, evidence specific to newly diagnosed pediatric AML‐M2 remains limited. Therefore, the present secondary analysis compared remission outcomes between low‐dose Ara‐C plus G‐CSF and standard‐dose induction chemotherapy in children with AML‐M2 and further evaluated survival, treatment‐related burden, and molecular subgroup outcomes.

2. Patients and Methods

2.1. Patients

The present study was a retrospective secondary analysis of prospectively collected data from the CALS III‐AML18 trial, a multicenter, randomized, Phase III noninferiority trial conducted at participating medical centers in China and registered with the Chinese Clinical Trial Registry (ChiCTR1800015883). For the present secondary analysis, patients with newly diagnosed AML‐M2 who were younger than 18 years and had not previously received chemotherapy were identified from 11 participating centers between June 2018 and June 2022. Patients retained their original randomized treatment assignments from the parent CALS III‐AML18 trial to either the low‐dose chemotherapy (LDC) arm or the standard‐dose chemotherapy (SDC) arm (Figure S1). Detailed eligibility criteria and randomization procedures of the parent CALS III‐AML18 trial have been reported previously [7].

2.2. Treatment Protocol

All patients received 2 courses of induction chemotherapy. The LDC arm received mitoxantrone at a daily dose of 5 mg/m² (Day 1, 3, 5), Ara‐C at a daily dose of 20 mg/m² (Day 1–10), and G‐CSF at a daily dose of 5 μg/kg (Day 1–10). The SDC arm was treated with daunorubicin at a daily dose of 50 mg/m² (Day 2, 4, 6), Ara‐C at a daily dose of 200 mg/m² (Day 1–10), and etoposide at a daily dose of 100 mg/m² (Day 1–5). Patients who did not achieve CR after induction chemotherapy received salvage chemotherapy. After CR, they were treated with an Ara‐C‐based consolidation regimen. Patients in the high‐risk group who were not candidates for hematopoietic stem‐cell transplantation (HSCT) were administered another Ara‐C cycle. Detailed treatment protocol is shown in Figure S2 and Table S1.

2.3. Genetic Detection

Pretreatment bone marrow aspirates were collected from enrolled patients. Genomic profiling was performed using next‐generation sequencing to assess mutations associated with AML. In addition, fluorescence in situ hybridization and multiplex reverse‐transcription polymerase chain reaction were used to detect cytogenetic abnormalities and fusion genes.

2.4. Assessments and End Points

Patients were classified into low‐, intermediate‐, and high‐risk groups according to the risk‐stratification criteria of the CALS III‐AML18 protocol, as detailed in Table S2. The principal response outcome was the CR and CR with incomplete recovery (CRi) rate after induction I and II. CR was defined as bone marrow blasts < 5%, absence of blasts with Auer rods, absence of extramedullary disease, absolute neutrophil count > 1.0 × 10⁹/L, platelet count > 80 × 10⁹/L, and independence of red cell transfusions (hemoglobin > 80 g/L). CRi was defined as fulfillment of the morphologic CR criteria without complete neutrophil and/or platelet recovery. Partial response (PR) was defined as a ≥ 50% reduction in bone marrow blasts, with residual blasts of 5%–19%, whereas no remission (NR) was defined as ≥ 20% bone marrow blasts. Relapse was defined as bone marrow blasts ≥ 5% (measurable residual disease (MRD) > 5%), the reappearance of blasts in the blood, or the development of extramedullary disease. Additional outcomes included overall survival (OS), defined as the time from treatment initiation to death from any cause; event‐free survival (EFS), defined as the time from treatment initiation to no response, relapse, or death; and relapse‐free survival (RFS), defined as the time from treatment initiation to induction treatment failure or relapse from CR/CRi. Adverse events were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 5.0, with grade ≥ 3 events classified as severe. Hematologic recovery and supportive‐care requirements during induction were also evaluated. Neutrophil recovery was defined as the time to an absolute neutrophil count of ≥ 0.5 × 10⁹/L, and platelet recovery as the time to a platelet count of ≥ 20 × 10⁹/L.

2.5. Statistical Analyses

Time‐to‐event outcomes were estimated using the Kaplan–Meier method and compared using log‐rank tests. Associations with OS, RFS, and EFS were evaluated using univariable and multivariable Cox proportional hazards regression models. Covariates were selected based on clinical relevance and included demographic and baseline clinical factors (sex, age, and white blood cell count), treatment assignment, AML1‐ETO fusion status, and recurrent gene mutations (KIT, CEBPA, ASXL1, ASXL2, and FLT3). The proportional hazards assumption was assessed by testing time‐dependent covariate interactions in the Cox models, with no evidence of violation observed. Cox regression results are reported as hazard ratios (HRs) with 95% confidence intervals (CIs). Response rates were compared using the χ ² test or Fisher's exact test, as appropriate. Factors associated with CR/CRi were evaluated using logistic regression and reported as odds ratios (ORs) with 95% CIs. Risk ratios (RRs) and 95% CIs were calculated using the Koopman asymptotic score method (GraphPad Prism 10.0). All tests were two‐sided with a significance threshold of 0.05. Statistical analyses were performed using SPSS (IBM, version 26.0, Chicago, IL).

3. Results

3.1. General Characteristic

From June 2018 to June 2022, 248 patients were screened, of whom 27 were excluded according to the parent‐trial criteria, leaving 221 patients for the analysis (Figure S3). Among the 221 patients included in this secondary analysis, 109 had been randomized to the LDC arm and 112 to the SDC arm in the parent CALS III‐AML18 trial. Baseline demographic and disease characteristics were generally comparable between the two arms (Table 1). The cohort included 135 males (61.1%) and 86 females (38.9%), with a median age of 8.24 (0.57–17.67) years. One patient was classified as low risk, and all others were categorized as intermediate‐ or high‐risk. The AML1‐ETO fusion gene resulting from t(8;21)(q22;q22) was detected in 63.3% of patients. Mutations in KIT, CEBPA, ASXL1, ASXL2, and FLT3 were detected in 83 (37.6%), 35 (15.8%), 27 (12.2%), 30 (13.6%), and 31 (14.0%) patients, respectively; among the 31 patients with FLT3 mutations, 22 had FLT3 internal tandem duplications (FLT3‐ITD) and 9 had FLT3 tyrosine kinase domain mutations (FLT3‐TKD). No statistically significant between‐group differences were observed in these baseline characteristics.

TABLE 1.

General characteristics.

Characteristic	Total (n = 221)	LDC arm (n = 109)	SDC arm (n = 112)	p
Sex
Male	135	61	74	0.123
Female	86	48	38
Age, years
0 ~ < 5	56	25	31	0.057
5 ~ < 10	86	51	35
10 ~ < 18	79	33	46
Risk group
Low risk	1	1	0	0.155
Intermediate risk	123	55	68
High risk	97	53	44
Fusion gene
t(8;21)/AML1‐ETO	140	65	75	0.258
inv.(16)/t(16;16); CBFB‐MYH11	2	2	0	0.242
t(9;11); MLL‐AF9	2	2	0	0.242
t(x;11); MLL‐r	16	9	7	0.565
Negative	39	23	16	0.184
Others	11	5	6	0.792
Abnormal karyotype	9	3	6	0.499
Unknown	2	0	2	0.498

Abbreviations: LDC, low‐dose chemotherapy; SDC, standard‐dose chemotherapy.

3.2. Treatment Response

At the end of induction I, the CR/CRi rates were 67.0% in the LDC arm and 70.5% in the SDC arm (p = 0.568). At the end of induction II, the corresponding rates were 84.4% and 89.3%, respectively (p = 0.740) (Table 2). After adjustment for sex, age, initial white blood cell count, cytogenetics, and risk group, treatment assignment was not associated with CR/CRi after either induction course (odds ratio = 1.235, p = 0.485; odds ratio = 1.524, p = 0.254, respectively) (Figures 1 and 2).

TABLE 2.

Response to the induction therapy.

Response	Total	LDC arm	SDC arm	p
Induction I	221	109	112	0.568
CR/CRi	152	73	79
PR	45	21	24
NR	19	13	6
Died	5	2	3
Induction II	192	92	100	0.740
CR/CRi	183	87	96
PR	7	4	3
NR	1	1	0
Died	1	0	1

Abbreviations: CR, complete remission; CRi, CR with incomplete recovery; LDC, low‐dose chemotherapy; NR, no remission; PR, partial remission; SDC, standard‐dose chemotherapy.

FIGURE 1.

Risk ratios for CR/CRi according to patient characteristics after induction I. CI, confidence interval; CR, complete remission; CRi, CR with incomplete recovery; HR, high risk; IR, intermediate risk; LDC, low‐dose chemotherapy; LR, low risk; RR, risk ratio; SDC, standard‐dose chemotherapy; WBC, White blood cell.

FIGURE 2.

Risk ratios for CR/CRi according to patient characteristics after induction II. CI, confidence interval; CR, complete remission; CRi, CR with incomplete recovery; HR, high risk; IR, intermediate risk; LDC, low‐dose chemotherapy; LR, low risk; RR, risk ratio; SDC, standard‐dose chemotherapy; WBC, white blood cell.

3.3. Survival Analysis

Overall, 178 of the 221 patients proceeded to consolidation treatment, including 87 in the LDC arm and 91 in the SDC arm (Figure S3). A total of 104 patients (47.1%) underwent HSCT, with a balanced distribution between the two arms (Table 1). The median follow‐up duration was 3.44 years. During this period, 33 patients died—19 (17.8%) in the LDC arm and 14 (12.7%) in the SDC arm (p = 0.302). Causes of death included HSCT‐related complications (n = 9, 1 of whom received a second HSCT), multiple organ failure due to disease progression (n = 16), and severe infection (n = 8). The 3 year OS rates were 81.2% ± 3.9% and 86.9% ± 3.3% in the LDC and SDC arms, respectively (p = 0.304). The 3 year RFS rates were 82.0% ± 3.8% and 88.1% ± 3.1%, respectively (p = 0.205), and the 3 year EFS rates were 62.1% ± 4.8% and 70.8% ± 4.4%, respectively (p = 0.218) (Figure 3 and Table S3).

FIGURE 3.

Survival analysis. (A) The 3 year OS was 81.2% ± 3.9% in the LDC arm and 86.9% ± 3.3% in the SDC arm (p = 0.304); (B) The 3 year RFS was 82.0% ± 3.8% in the LDC arm and 88.1% ± 3.1% in the SDC arm (p = 0.205); (C) The 3 year EFS was 62.1% ± 4.8% in the LDC arm and 70.8% ± 4.4% in the SDC arm (p = 0.218). EFS, event‐free survival; LDC, low‐dose chemotherapy; OS, overall survival, RFS, relapse‐free survival; SDC, standard‐dose chemotherapy.

Further analyses explored survival outcomes according to molecular characteristics (Table S3; Figures [Link], [Link]). Among patients with KIT mutations (34 in the LDC group and 49 in the SDC group), survival outcomes were significantly poorer in the LDC arm than in the SDC arm, with 3 year OS rates of 67.1% ± 8.7% versus 91.5% ± 4.1% (p = 0.011), RFS rates of 70.6% ± 7.8% versus 87.5% ± 4.8% (p = 0.038), and EFS rates of 49.2% ± 8.7% versus 81.0% ± 5.7% (p = 0.002). Within the LDC arm, KIT‐mutated patients also had poorer OS, RFS, and EFS than patients without KIT mutations (p = 0.031, 0.016, and 0.035, respectively), whereas these differences were not statistically significant within the SDC arm (p = 0.224, 0.863, and 0.071, respectively).

In multivariable Cox analyses, treatment assignment was not significantly associated with OS, RFS, or EFS, whereas KIT and ASXL1 mutations were associated with increased mortality risk (Tables S4 and S5). Importantly, unlike KIT, ASXL1 mutation was not associated with a significant difference in survival between the LDC and SDC arms in the treatment‐specific subgroup analysis (Table S3). Among patients with AML1‐ETO, however, those with ASXL1 mutations had poorer OS and EFS than those without ASXL1 mutations (3 year OS: 66.2% ± 11.3% vs. 88.7% ± 3.0%, p = 0.009; 3 year EFS: 50.0% ± 11.8% vs. 73.3% ± 4.1%, p = 0.020), whereas RFS did not differ significantly (p = 0.882).

For FLT3, patients with FLT3‐TKD showed numerically higher OS, RFS, and EFS than those with FLT3‐ITD, but none of these differences reached statistical significance (3 year OS: 88.9% ± 10.5% vs. 86.4% ± 7.3%, p = 0.838; 3 year RFS: 100% vs. 85.6% ± 7.8%, p = 0.275; 3 year EFS: 77.8% ± 13.9% vs. 48.5% ± 11.0%, p = 0.190).

3.4. Adverse Events and Treatment‐Related Burden

Sepsis was the most frequently reported adverse event, occurring in 63 patients (30 in the LDC arm and 33 in the SDC arm) (Table 3). Among patients who developed sepsis, Grade 3–5 events occurred in 10 of 30 patients (33.3%) in the LDC arm and 26 of 33 patients (78.8%) in the SDC arm (p < 0.001). Seventeen patients developed septic shock (6 in the LDC arm and 11 in the SDC arm), and 13 developed severe pneumonia (9 and 4 patients, respectively); neither difference was statistically significant. Intracranial hemorrhage and cardiac insufficiency were uncommon, with no significant between‐group differences (Table 3). Eight patients with septic shock died, including 1 in the LDC arm and 7 in the SDC arm. Hematologic toxicities were frequent in both groups. Neutropenia occurred in 101 of 109 patients in the LDC arm and 109 of 112 patients in the SDC arm (p = 0.111), anemia in 92 and 101 patients, respectively (p = 0.197), and thrombocytopenia in 99 and 109 patients, respectively (p = 0.040). No second primary malignancies were observed during follow‐up.

TABLE 3.

Adverse event.

	Total	LDC arm	SDC arm	p
Sepsis	63	30	33	0.749
Grade 1 ~ 2	27	20	7	< 0.001*
Grade 3 ~ 5	36	10	26
Septic Shock	17	6	11	0.314
Severe pneumonia	13	9	4	0.162
Intracranial hemorrhage	2	0	2	0.498
Cardiac insufficiency	3	2	1	0.618
Hematologic
Neutropenia	210	101	109	0.111
Thrombocytopenia	208	99	109	0.040*
Anemia	193	92	101	0.197

Abbreviations: LDC, low‐dose chemotherapy; SDC, standard‐dose chemotherapy.

^* p < 0.05.

Hematologic recovery and supportive‐care requirements are summarized in Table S6. Compared with the SDC arm, patients in the LDC arm had significantly shorter neutrophil and platelet recovery times during both induction courses, together with lower red blood cell and platelet transfusion requirements and lower treatment costs (Table S6).

4. Discussion

Our study focused on children with AML‐M2 derived from the CALS III‐AML18 trial. After two induction courses, CR/CRi was achieved in 84.4% of patients receiving LDC and 89.3% receiving SDC, with no significant difference between the two groups. Three‐year OS, RFS, and EFS were likewise similar between treatment arms. The main distinction was observed in treatment‐related toxicity, with fewer severe septic events and less thrombocytopenia in the LDC group. These findings suggest that reducing induction intensity did not result in an obvious loss of disease control in the overall AML‐M2 cohort, although outcomes differed in some molecular subgroups.

The remission rate observed in our AML‐M2 cohort is consistent with previous experience using reduced‐intensity induction in pediatric AML. In an earlier pediatric observational study, LDC produced a CR rate of 88.8%, compared with 86.4% with SDC, with similar 5 year OS (72.7% vs. 72.5%) and EFS (61.4% vs. 65.2%) [6]. More recently, the parent CALS III‐AML18 randomized trial reported CR/CRi rates of 95.1% and 95.3% with LDC and SDC, respectively, together with similar 4 year OS and EFS [7]. The somewhat lower remission rates in the present analysis may reflect the restriction to AML‐M2 and its distinct molecular composition. G‐CSF may contribute to the activity of the low‐dose regimen. Previous studies have shown that G‐CSF priming can alter the response of AML blasts to chemotherapy [10]. Nomdedeu et al. [11] suggested that G‐CSF may reduce stromal protection of leukemic cells and impair their clonogenic survival. However, clinical studies of G‐CSF administration have yielded variable results, indicating that its effect is likely dependent on disease context and the accompanying chemotherapy regimen [12, 13].

Treatment‐related toxicity is particularly relevant in pediatric AML because intensive induction is accompanied by prolonged myelosuppression and a high burden of infectious complications [14]. In our cohort, severe sepsis was observed less often with LDC, and thrombocytopenia was also less frequent, whereas the occurrence of neutropenia and anemia was similar between the two groups. In addition, LDC was associated with faster neutrophil and platelet recovery during both induction courses, lower transfusion requirements, and reduced treatment costs. Taken together, these findings suggest that the potential clinical advantage of LDC may lie primarily in faster hematologic recovery and a lower treatment‐related burden.

The molecular findings further indicate that AML‐M2 should not be considered a biologically uniform group when induction intensity is reduced. KIT‐mutated patients treated with LDC had inferior OS, RFS, and EFS compared with those receiving SDC, and KIT and ASXL1 mutations were also associated with poorer prognosis in the overall cohort. This observation is consistent with genomic studies of pediatric AML showing that KIT mutations, particularly exon 17 variants in AML1‐ETO‐positive AML, identify patients with less favorable outcomes [15]. A recent multicenter study of AML1‐ETO–positive AML similarly showed that combining KIT/ASXL1 status with MRD markedly improved risk stratification, with substantially higher relapse and mortality in the high‐risk group [16]. These data support retaining molecular information when considering any reduction in chemotherapy intensity. Our FLT3‐ITD subgroup also showed numerically poorer outcomes than FLT3‐TKD, although the difference was not statistically significant. Mansor et al. [17] reported associations of TREM1 overexpression with FLT3‐ITD, CD123 expression, chemotherapy resistance, and inferior survival, raising the possibility that inflammatory and stemness‐related signaling contributes to the adverse biology of FLT3‐ITD AML. Whether a similar relationship is present in pediatric AML‐M2 requires further study.

Clinically, these findings suggest that LDC may warrant consideration as a toxicity‐sparing induction approach in selected children with AML‐M2 when treatment‐related toxicity is a major concern. However, the poorer outcomes observed with LDC among patients with KIT mutations support caution when considering this regimen in this molecular subgroup.

Several limitations should be acknowledged. The present study was a secondary analysis of the AML‐M2 subgroup from the randomized CALS III‐AML18 trial and was not independently powered for AML‐M2‐specific comparisons. Although the original randomized treatment assignments were retained unchanged, restricting the analysis to trial‐eligible patients with AML‐M2 may limit the generalizability of the findings to the broader pediatric AML‐M2 population. In addition, the molecular subgroup analyses, particularly those involving KIT, ASXL1, and other less frequent alterations, were limited by small sample sizes and should therefore be interpreted cautiously. The current follow‐up duration may also limit assessment of late relapse, secondary malignancies, and other late treatment‐related complications.

Future prospective studies with larger AML‐M2 cohorts, adequate representation of molecular subgroups, and longer follow‐up are needed to confirm the durability of these findings and determine whether molecular features can reliably inform selection of induction strategies.

In conclusion, no statistically significant differences in remission or survival outcomes were observed between LDC and SDC in this AML‐M2 cohort, while LDC was associated with fewer severe septic events, faster hematologic recovery, and lower supportive‐care requirements. These findings support further evaluation of LDC as a toxicity‐sparing induction strategy in selected pediatric patients with AML‐M2. The poorer outcomes among patients with KIT mutations suggest that molecular features should be considered when treatment intensity is reduced.

Author Contributions

Wenxin Ou: writing – original draft, data curation. Xiaohua Zhu: formal analysis. Jun Lu: visualization, writing – review and editing. Hongsheng Wang: visualization, writing – review and editing. Ning Liao: writing – review and editing. Yufeng Liu: writing – review and editing. Hua Jiang: writing – review and editing. Qi An: writing – review and editing. Xiaojun Yuan: writing – review and editing. Hailong He: writing – review and editing. Ningling Wang: formal analysis. Jixia Luo: formal analysis. Xiuli Ju: writing – review and editing. Qian‐Fei Wang: writing – review and editing. Minghua Yang: data curation. Cheng Cheng: data curation. Peifang Xiao: data curation. Liangchun Yang: data curation. Li Gao: formal analysis. Raul C. Ribeiro: writing – review and editing. Shaoyan Hu: project administration. Xiaowen Zhai: project administration.

Funding

This work was supported by the National Natural Science Foundation of China, 82141125. Science and Technology Commission of Shanghai Municipality, 21Y31900302, 24Y12800602. Shanghai Hospital Development Center, SHDC12019121. National Key R&D Program of China, 2022YFC2705003, 2023YFC2706301.

Ethics Statement

This study was conducted in accordance with the Declaration of Helsinki and has been approved by the ethics committee and clinical registration (ChiCTR1800015883).

Consent

Informed consent was obtained from all patients or legal guardians.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: AML protocol outline.

Figure S2: Induction therapy protocol.

Figure S3: Patient distribution flowchart.

Figure S4: Survival analysis curves of the patients in each subgroup. A, B and C represent the OS, RFS, and EFS, respectively, of patients with AML1‐ETO fusion gene divided into the LDC arm and the SDC arm.

Figure S5: Survival analysis curves of the patients in each subgroup. A, B and C represent the OS, RFS, and EFS, respectively, of patients with mutated KIT divided into the LDC arm and the SDC arm.

Figure S6: Survival analysis curves of the patients in each subgroup. A, B and C represent the OS, RFS, and EFS, respectively, of patients with mutated CEBPA divided into the LDC arm and the SDC arm.

Figure S7: Survival analysis curves of the patients in each subgroup. A, B and C represent the OS, RFS, and EFS, respectively, of patients with mutated ASXL1 divided into the LDC arm and the SDC arm.

Figure S8: Survival analysis curves of the patients in each subgroup. A, B and C represent the OS, RFS, and EFS, respectively, of patients with mutated ASXL2 divided into the LDC arm and the SDC arm.

Figure S9: Survival analysis curves of the patients in each subgroup. A, B and C represent the OS, RFS, and EFS, respectively, of patients with mutated FLT3 divided into the LDC arm and the SDC arm.

Table S1: Treatment regimen.

Table S2: Risk stratification.

Table S3: Survival analysis.

Table S4: Univariable cox regression analyses for OS, RFS, and EFS.

Table S5: Multivariable cox regression analyses for OS, RFS, and EFS.

Table S6: Hematologic recovery, transfusion requirements, and treatment costs by induction regimen.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.