Clinical characteristics and therapeutic determinants of RUNX1::RUNX1T1 differ from those of CBFB::MYH11 acute myeloid leukemia

Abstract

Core binding factor acute myeloid leukemia (CBF-AML) includes RUNX1::RUNX1T1 and CBFB::MYH11 AML. To investigate whether they should be regarded as distinct entities and treated separately, we retrospectively analyzed 536 patients with CBFAML aged 60 years or younger. For CBFB::MYH11 AML, no outcome differences were observed between standard-dose (SD) and intermediate-dose (ID) cytarabine induction, with 5-year overall survival (OS) and relapse-free survival (RFS) at 86.4% versus 85.3% (P=0.99) and 74.1% versus 68.4% (P=0.93), respectively. However, ID induction improved the outcomes of RUNX- 1::RUNX1T1 AML, with 5-year OS and RFS rates of 77.7% versus 60.3% (P<0.001) and 71.4% versus 54.1% (P<0.001), respectively. Allogeneic hematopoietic stem cell transplantation (allo-HSCT) improved CBFB::MYH11 patients’ RFS but not OS for both the entire CBFB::MYH11 cohort and patients with a measurable residual disease (MRD) <3-log reduction after two courses of chemotherapy. Allo-HSCT improved survival of RUNX1::RUNX1T1 patients with MRD <3-log reduction when receiving SD induction, with 5-year OS and RFS rates for transplantation versus non-transplantation of 71.9% versus 43.6% (P=0.023) and 72.9 % versus 35.0% (P=0.001), while conferring no additional benefits for those with ID regimen, with a 5-year OS and RFS of 35.6% versus 71.2% (P=0.73) and 37.4% versus 60.7% (P=0.71), respectively. Overall, our study suggests that RUNX1::RUNX1T1 and CBFB::MYH11 AML are different entities and should be treated with distinct strategies.

Introduction

Core binding factor acute myeloid leukemia (CBF-AML) is a common type of leukemia, accounting for approximately 15% of de novo adult AML.^1,2 It is characterized by the formation of RUNX1::RUNX1T1 or CBFB::MYH11 fusion genes, corresponding to t(8;21) or inv(16) chromosomal rearrangements. The two chromosomal rearrangements disrupt the subunits encoding the CBF complex, thereby impairing normal hematopoiesis and leading to maturation arrest.³

Despite being classified as distinct entities in current AML classification systems, RUNX1::RUNX1T1 and CBF-B::MYH11-positive AML are frequently grouped and investigated collectively in clinical research due to their overlapping pathogenic mechanisms and similarly favorable prognostic profiles.⁴ These two types of AML share similar treatment regimens and strategies. The ‘7+3’ regimen (7 days of continuous infusion cytarabine plus 3 days of anthracycline) with or without gemtuzumab ozogamicin (GO) serving as the standard induction regimen is recommended by current guidelines for both genetic groups.^5,6 However, patients with RUNX1::RUNX1T1 or CBFB::MYH11 AML seem to display notable differences in several biological features. For instance, the mutation profiles revealed that RUNX1::RUNX1T1 patients frequently harbor mutations in epigenetic regulators, chromatin modulators, or cohesion proteins, whereas CBFB::MYH11 AML is more commonly associated with kinase mutations.^7-9 Although both RUNX-1::RUNX1T1 and CBFB::MYH11 AML are categorized as favorable risk according to European LeukemiaNet (ELN) risk stratification,⁵ their outcomes diverge significantly. Overall, patients with CBFB::MYH11 have a lower risk of relapse and longer survival compared to those with RUNX1::RUNX1T1.^10,11 Hence, the question arises as to whether patients with RUNX1::RUNX1T1 and CBFB::MYH11 AML should be regarded as distinct entities based not only on biological differences, but also on therapeutic determinants. To answer this question, we conducted a retrospective analysis of 370 patients with RUNX1::RUNX1T1 and 166 with CBFB::MYH11 at our center with the aim of elucidating the differential clinical characteristics and therapeutic determinants of these two AML subtypes.

Methods

Patients

In total, 3,144 consecutive patients with newly diagnosed AML at our center between January 2011 and January 2024 were retrospectively assessed in this study. Enrolled patients had to meet the following criteria: (i) onset age not exceeding 60 years, (ii) CBF-AML, and (iii) receiving intensive inductions. This study represents an expanded and updated cohort of patients with CBF-AML, building on our previously published work.¹² The study was approved by the Blood Diseases Hospital Ethics Committee and was conducted in accordance with the Declaration of Helsinki. Written informed consent for treatment and biological sample collection was obtained from all patients.

Treatments

All patients received intensive induction therapy with a combination of standard-dose (SD) (100 mg/m²/day 1-7 as a 12-hour [h] intravenous [iv] infusion) or intermediate-dose (ID) cytarabine (100 mg/m²/day 1-4 as a 12-h iv infusion and 1 mg/m²/day every 12 h as a 3-h iv infusion on days 5-7) plus anthracycline ± homoharringtonine (HHT), followed by three to four courses of consolidation regimens with high or intermediate doses of cytarabine after complete remission (CR). The treatment regimen was consistent with the therapeutic protocols previously reported for AML patients in our center.^13,14 Intrathecal injections were administered for routine prophylaxis of central nervous system leukemia. The regimen typically included cytarabine (40-50 mg/dose), methotrexate (5-15 mg/dose), and dexamethasone (5-10 mg/dose), with two doses administered during induction and consolidation phases. Hematopoietic stem cell transplantation (HSCT) was recommended for patients with a measurable residual disease (MRD) <3-log reduction after two courses of chemotherapy. In our study, 86 patients underwent HSCT during first complete remission (CR1). Of these, 69 (80.2%) had MRD-based indications for transplantation, while 17 (19.8%) opted for HSCT despite achieving >3-log MRD reduction post-induction.

Next-generation sequencing

Next-generation sequencing (NGS) was performed on a custom targeted gene panel of 267 genes closely related to hematological neoplasms through the Illumina platform with a sensitivity of 1%. Total DNA was extracted from the bone marrow mononuclear cells. Using ribosomal RNA depletion for library preparation, high-throughput sequencing was performed at a depth of ranging from 1,000X to 2,000X. The visualization of mutations was performed by complexheatmap R package.¹⁵ NGS was conducted on 331 patients, representing 61.8% of the total cohort.

Measurable residual disease detection by quantitative polymerase chain reaction

Fresh bone marrow samples were collected, and total RNA was isolated using an RNA extraction kit following the manufacturer’s protocol. Extracted RNA was analyzed by real-time quantitative polymerase chain reaction (qPCR) for RUNX1::RUNX1T1 and CBFB::MYH11 fusion transcripts on an ABI QuanStudio 5. Values were expressed as a percentage of the fusion transcript to normalizing ABL1. The sensitivity of detection was 1 in 100,000. The patients underwent fusion transcript testing at diagnosis and after each course. The MRD level was calculated as the log reduction in transcript levels between diagnosis and two cycles of chemotherapy. In total, 512 patients underwent serial MRD assessments, with 359 (97.0%) patients in the RUNX1::RUNX1T1 group and 153 (92.2%) patients in the CBFB::MYH11 group.

Statistical analysis

Descriptive statistics were used to describe patients’ baseline characteristics, and data comparisons were conducted using the χ² test or Mann-Whitney U test. Cumulative incidence of relapse (CIR) was measured from the date of CR until the date of relapse or last follow-up. CIR was compared by Fine-Gray test. Overall survival (OS) was defined as the interval from the date of diagnosis to death from any cause or last follow-up. Relapse-free survival (RFS) was defined as the time from achieving complete remission (CR) to the first relapse, death, or the date of the last follow-up. OS and RFS were evaluated by Kaplan-Meier method and compared by the Log-rank test. Landmark analysis was used to avoid bias introduced by patients who experienced early relapse or death before HSCT when analyzing the effect of HSCT. The landmark day was set as the median time from CR to HSCT. Statistical significance was set at P<0.05. This study applied R version 4.2.0 for graphing and statistical analysis.

Results

Clinical characteristics

A total of 536 patients with newly diagnosed CBF-AML were retrospectively enrolled and treated between January 2011 and January 2024 at our center. Figure 1 depicts the detailed screening process. Among the 536 patients, 303 (56.5%) were male and 233 (43.5%) were female, with a median age of onset at 35 years (range, 14-60 years). According to the World Health Organization (WHO) 2022 classification,¹⁶ 370 (69.0%) cases were categorized as AML with RUNX1::RUNX1T1 and 166 (31.0%) as AML with CBFB::MYH11. Patients with RUNX1::RUNX1T1 presented with an earlier age of onset and lower white blood cell counts (WBC), with a median age of 34 years versus 39.5 years (P<0.001) and a median WBC count of 8.5 versus 30.3 (P<0.001) than patients with CBFB::MYH11, respectively. All patients in the cohort were treated with intensive chemotherapy, with 185 (50.0%) of RUNX1::RUNX1T1 and 99 (59.6%) of CBFB::MYH11 undergoing induction with SD cytarabine, and 185 (50.0%) and 67 (40.4%) receiving induction with ID cytarabine, respectively. Additionally, 60 (16.2%) patients with RUNX1::RUNX1T1 and 26 (15.7%) with CBFB::MYH11 underwent HSCT in CR1. The detailed clinical characteristics of the two groups were presented in Table 1.

Figure 1.Study flow diagram. ChiCTR-TRC-10001202: a phase III study on optimizing treatment based on risk stratification for acute myeloid leukemia; clincaltrials gov. Identifier: NCT03021330: efficacy of intermediate-dose cytarabine induction regimen in adult acute myeloid leukemia (AML); clincaltrials gov. Identifier: NCT02432872: treatment of older adult acute myeloid leukemia patients aged 55 to 65 years. CBF-AML: core-binding factor acute myeloid leukemia; MRD: measurable residual disease.

Gene mutation profiles and cytogenetic abnormalities

Genetic mutation spectrum revealed that both subtypes of AML were frequently associated with kinase-related gene mutations, such as KIT, RAS, and FLT3. In contrast, genes involved in epigenetic regulation, such as ASXL1, ASXL2, TET2, and ZBTB7A, were more likely to be found in patients with RUNX1::RUNX1T1 compared to those with CBFB::MYH11 (Online Supplementary Figure S1). We then investigated whether gene mutations in a sample size of more than ten patients affected patients’ survival. Univariate analysis revealed that, for patients with RUNX1::RUNX1T1, the presence of KIT and TET2 mutations adversely correlated with OS and RFS, with hazard ratio (HR) for OS of 2.81 (95% confidence interval [CI]: 1.48-5.33; P=0.002) and 3.80 (95% CI: 1.59-9.09; P=0.003), and for RFS at 2.86 (95% CI: 1.69-4.82; P<0.001) and 3.57 (95% CI: 1.75-7.29; P<0.001), respectively. While NRAS mutations were generally indicative of better OS and RFS, with HR=0.31 (95% CI: 0.11-0.88; P=0.027) for OS and 0.38 (95% CI: 0.17-0.83; P=0.015) for RFS. In contrast, these genetic aberrations did not exert a statistically significant impact on the survival of patients with CBFB::MYH11 (Online Supplementary Table S1). Online Supplementary Figure S2 presented the survival curves for patients with and without the specified gene mutations in RUNX1::RUNX1T1 AML. Other gene mutations have not been further explored due to insufficient sample sizes.

We next compared the cytogenetic profiles between the two CBF-AML subtypes. As summarized in Table 1, loss of sex chromosomes (52.9%) was the most prevalent abnormality in RUNX1::RUNX1T1 patients, followed by deletion of chromosome 9 (5.8%). In CBFB::MYH11 patients, the predominant cytogenetic abnormalities were trisomy 22 (23.3%) and trisomy 8 (8.8%). To evaluate their prognostic significance, we conducted survival analyses focusing on these recurrent chromosomal alterations. In RUNX1::RUNX1T1 patients, neither sex chromosome loss nor chromosome 9 deletion significantly impacted survival outcomes. Patients with sex chromosome loss showed comparable OS (5-year OS: 72.4±3.8% vs. 67.1±4.1%; P=0.35) and RFS (5-year RFS: 66.2±3.8% vs. 60.8±4.2%; P=0.61) to those without this abnormality. Similarly, deletion of chromosome 9 did not significantly affect outcomes, with 5-year OS and RFS at 82.1±9.4% versus 69.1±2.9% (P=0.29) and 73.8±10.2% versus 62.1±3.0% (P=0.38) (Online Supplementary Figure S3A-D). Among CBFB::MYH11 patients, trisomy 8 showed a trend toward better RFS at 100% versus 69.0±5.3% (P=0.082), while OS remained comparable between groups at 100% versus 85.7±3.4% (P=0.21) (Online Supplementary Figure S4A, B). Trisomy 22 status did not significantly influence survival outcomes, with similar OS (91.7±5.7% vs. 85.3±3.7%; P=0.23) and RFS (70.1±5.7% vs. 76.5±9.9%; P=0.31) observed regardless of its presence (Online Supplementary Figure S4C, D). Other chromosomal abnormalities were not further analyzed due to limited sample size.

Table 1.Clinical characteristics of the patients diagnosed as core binding factor acute myeloid leukemia.

Response and outcomes

Among the 536 patients, 513 (95.7%) achieved CR within two cycles of induction, including 353 (95.4%) with RUNX-1::RUNX1T1 and 160 (96.4%) with CBFB::MYH11. A total of 512 in these 513 CR patients had informative MRD testing results. Of these, 226 (63.0%) in RUNX1::RUNX1T1 and 43 (28.1%) in CBFB::MYH11 achieved a >3-log MRD reduction, which included 66 (18.4%) and 11 (7.2%) patients, respectively, who demonstrated PCR-undetectable MRD. In the RUNX1::RUNX1T1 group, 131 patients (72.4%) in the ID cytarabine induction achieved a >3-log MRD reduction, compared to 95 patients (53.4%) in the SD group, while in the CBFB::MYH11 group, 23 patients (35.9%) in the ID group achieved this reduction versus 20 patients (22.5%) in the SD group. With a median follow-up of survivors of 50 months, the estimated 5-year OS and RFS were 70.0±2.8% versus 86.1±3.3% (P=0.003) and 63.7±2.8% versus 71.1±4.9% (P=0.010) for RUNX1::RUNX1T1 and CBFB::MYH11 patients, respectively (Figure 2A, B). The 5-year OS and RFS censored at HSCT were 71.2±2.9% versus 85.7± 3.6% (P=0.010) and 65.7±2.9% versus 66.7±5.8% (P=0.042), respectively (Online Supplementary Figure S5). We then performed a multivariate analysis incorporating age, sex, WBC, induction regimen, CR1 HSCT, and gene mutations in KIT, TET2, and NRAS. The results indicated that for patients with RUNX1::RUNX1T1, age, induction regimen, and mutations in KIT, TET2, and NRAS served as independent prognostic factors for OS, but not for patients with CBFB::MYH11 (Table 2).

Figure 2.Outcomes of patients with RUNX1::RUNX1T1 and CBFB::MYH11. (A) Overall survival (OS) and (B) relapse-free survival (RFS). The red line represents patients with CBFB::MYH11, whereas the blue line represents those with RUNX1::RUNX1T1.

Table 2.Multivariate analysis results for overall survival of patients with core binding factor acute myeloid leukemia.

Impact of induction regimen

We explored the effects of the induction regimen on the patients’ outcomes. For RUNX1::RUNX1T1 patients, the ID regimen demonstrated a significant advantage over the SD regimen, with 5-year OS and RFS rates of 77.7±3.4% versus 60.3±4.5% (P<0.001) and 71.4±3.5% versus 54.1±4.6% (P<0.001), respectively. In contrast, for CBFB::MYH11 patients, no significant differences in survival were observed between the two inductions, with 5-year OS and RFS at 85.3±5.0% versus 86.4±4.4% (P=0.99) and 68.4±7.7% versus 74.1± 5.8% (P=0.93) (Figure 3), respectively. Although all patients received high-intensity induction, variations in anthracycline doses and HHT use existed across trials.

To eliminate potential biases arising from these variations, we conducted separate analyses within two primary clinical trials. For RUNX1::RUNX1T1 patients, both clinical trials demonstrated that the ID group had better OS and RFS compared to the SD group (Online Supplementary Figure S6A-D), while CBFB::MYH11 patients exhibited no significant survival differences between ID and SD groups (Online Supplementary Figure S7A-D). Next, we investigated the effect of MRD levels on the outcomes of the two patient groups. It was found that patients with a <3-log MRD reduction in RUNX1::RUNX1T1 had significantly poorer outcomes than those with a >3-log MRD reduction, with 5-year CIR, OS, and RFS rates of 46.2±4.9% versus 26.5±3.2% (P<0.001), 55.3±5.2% versus 78.8±3.1% (P<0.001), and 47.9± 5.2% versus 71.7±3.3% (P<0.001) (Figure 4A, C and E), respectively. For CBFB::MYH11 AML, patients with a <3-log MRD reduction demonstrated a trend associated with higher CIR with 5-year CIR rates of 30.1±5.8% versus 14.9±8.9% (P=0.069); however, they shared similar OS and RFS with that of their counterparts, 5-year OS rates of 87.3±4.0% versus 96.8±3.2% (P=0.20), and 5-year RFS rates of 68.1±5.9% versus 81.8±9.0% (P=0.12) (Figure 4B, D and F). For CBFB::MYH11 AML, we did not find significantly different OS and RFS, even with a 2-log threshold (Online Supplementary Figure S8). We could not further decrease the threshold due to the limited sample size in the low-level MRD group.

Figure 3.Outcomes of patients with RUNX1::RUNX1T1 and CBFB::MYH11 in different induction regimens groups. (A) Overall survival (OS) and (B) relapse-free survival (RFS). The blue line represents RUNX1::RUNX1T1 patients receiving intermediate-dose (ID) induction, while the green line represents standard-dose (SD) regimen. The red line represents CBFB::MYH11 patients with the ID regimen, while sky blue represents the SD regimen.

Figure 4.Outcomes of patients with RUNX1::RUNX1T1 (A, C, and E) and CBFB::MYH11 (B, D, and F) in the different measurable residual disease groups. (A and B) Cumulative incidence of relapse (CIR); (C and D) overall survival (OS); and (E and F) are relapse-free survival (RFS). The red line represents patients with a <3-log measurable residual disease (MRD) reduction, while the blue line represents patients with a >3-log MRD.

Impact of hematopoietic stem cell transplantation in first complete remission

We further investigated the role of HSCT in CBF-AML patients. To avoid bias introduced by patients who experienced early relapse and death, a landmark analysis was performed. The landmark day was set at 7.2 months. HSCT improved CBFB::MYH11 patients’ RFS but not OS, with 5-year RFS at 91.6± 5.7% with HSCT, while 65.8±5.8% without HSCT (P=0.037), and 5-year OS at 91.6±5.7% versus 88.1±3.6%, respectively (P=0.69) (Online Supplementary Figure S9). When comparing survival of patients censored for HSCT, patients underwent HSCT still had improved 5-year RFS at 91.6±5.7%, while 66.7± 5.8% in patients censored for HSCT (P=0.045), while no significant difference in OS with 5-year OS at 91.6±5.7% and 85.7±3.6% (P=0.46), respectively (Online Supplementary Figure S10A, B). However, for all RUNX1::RUNX1T1 patients, HSCT did not confer a benefit, as the OS and RFS of patients with HSCT were comparable to those of non-HSCT or censored for HSCT (Online Supplementary Figured S9 and S10C, D). We then sought to determine whether HSCT could improve the outcomes in patients with high MRD levels. For patients with a <3-log MRD reduction, those who underwent HSCT demonstrated significantly improved RFS compared to non-transplanted patients, with the RUNX1::RUNX1T1 group showing 51.6±12.1% versus 46.3 ±6.0% (P=0.025) and the CBFB::MYH11 group showing 95.0±4.9% versus 60.9±7.0%, respectively (P=0.027). However, both groups only had a non-significant trend toward better OS, with rates of 52.6±16.8% versus 56.6± 6.1% (P=0.13) and 95.0±4.9% versus 85.3±4.9% (P=0.45), respectively. In RUNX1::RUNX1T1 patients with a >3-log MRD reduction, the transplantation group exhibited comparable OS and RFS to the non-transplantation group (P=0.56 and P=0.98, respectively) (Online Supplementary Figure S11). For CBFB::MYH11 patients with a >3-log MRD reduction, we could not compare the effects of HSCT on outcomes because only four patients with a >3-log MRD reduction received HSCT.

Given the finding that the ID induction regimen can improve outcomes for patients with RUNX1::RUNX1T1 compared to the SD regimen, we further investigated the benefits of HSCT in patients receiving different regimens with residual disease. As shown in Figure 5A-D, for patients who received SD induction, HSCT did provide additional benefits for those with a <3-log MRD reduction, with OS and RFS rates of 71.9±11.5% versus 43.6±9.1% (P=0.023) and 72.9±11.8% versus 35.0±8.2% (P=0.001) (Figure 5A, B), respectively. For patients with a >3-log MRD reduction, we could not investigate the benefits of HSCT because only five patients underwent HSCT (Figure 5C, D). Interestingly, among patients receiving ID induction, HSCT did not further improve survival for patients with a <3-log MRD reduction; the 5-year OS and RFS for those undergoing transplantation versus non-transplantation were 35.6±26.2% and 37.4±16.4%, 71.2±8.2% and 60.7±8.7%, respectively (P=0.73 and P=0.71) (Figure 5E, F). Only 15 RUNX1::RUNX1T1 patients receiving ID induction therapy demonstrated a <2-log MRD reduction. HSCT seemed to improve OS and RFS compared to chemotherapy, although no statistical differences were observed (Online Supplementary Figure S12).

Discussion

In this study, we found that ID cytarabine induction significantly improved outcomes for patients with RUNX-1::RUNX1T1, but not for patients with CBFB::MYH11. HSCT improved CBFB::MYH11 patients’ RFS but not OS for both the entire CBFB::MYH11 cohort and patients with a <3-log MRD reduction after two courses of chemotherapy. For patients with RUNX1::RUNX1T1 who received SD induction, HSCT did provide additional benefits for those with a <3-log MRD reduction after two courses of chemotherapy, while for those treated with ID cytarabine, HSCT did not confer additional benefits.

Although both RUNX1::RUNX1T1 and CBFB::MYH11 AML affect the function of the CBF complex, CBFB::MYH11 AML carries a better prognosis.¹¹ Consistent with previous studies, our research confirmed that, despite similar CR rates of over 90% after two induction cycles in both subtypes, the risk of relapse was lower and OS was better in patients with CBFB::MYH11 compared to those with RUNX1::RUNX1T1. Our study revealed an intriguing paradox: CBFB::MYH11 patients showed slower initial MRD clearance but superior outcomes compared to RUNX1::RUNX1T1 patients. Serial MRD data showed that while 63.0%, 71.3%, and 78.9% of RUNX1::RUNX1T1 patients achieved a >3-log reduction after two, three and four cycles of chemotherapy, respectively, CBFB::MYH11 patients showed more gradual but sustained clearance (28.1%, 46.3%, and 63.6%). This differential MRD kinetics may provide mechanistic insights into the superior clinical outcomes observed in CBFB::MYH11 patients. The sustained MRD reduction in this group likely reflects more effective eradication of leukemic clones, whereas the rapid but potentially incomplete clearance in RUNX1::RUNX1T1 patients may predispose to disease recurrence. Further-more, the higher prevalence of adverse-risk mutations in RUNX1::RUNX1T1 patients, may compound this effect by conferring intrinsic treatment resistance.

Figure 5.Outcomes of RUNX1::RUNX1T1 patients with or without hematopoietic stem cell transplantation in different induction regimens. (A-D) Standard-dose regimen (SD); (E-H) intermediate-dose regimen (ID) and measurable residual disease (MRD) levels: (A, B) and (E, F) <3-log; (C, D) and (G, H) >3-log. The red line represents patients who underwent hematopoietic stem cell transplantation (HSCT), whereas the blue line represents patients who did not undergo HSCT. OS: overall survival; RFS: relapse-free survival.

The optimal induction dose of cytarabine for AML has been the focus of extensive research, with studies yielding varying conclusions.¹⁷ While Willemze et al. found that high-dose cytarabine during induction produced higher remission and survival rates than SD cytarabine, especially in patients younger than age 46 years,¹⁸ Löwenberg et al. demonstrated that high-dose cytarabine resulted in excessive toxic effects without therapeutic benefits in patients aged 18 to 60 years.¹⁹ Similarly, another study showed no significant benefit of ID cytarabine over SD cytarabine.²⁰ In our study, we observed that ID cytarabine induction was associated with improved MRD response rates and survival outcomes in patients with RUNX1::RUNX1T1, while the benefit was less pronounced in CBFB::MYH11 patients. These findings align with the notion that the efficacy of cytarabine dose intensification may depend on the genetic context. However, the conflicting results across studies underscore the need for further research to identify the patient subgroups most likely to benefit from dose escalation.

MRD is a well-established prognostic factor of AML.²¹ Previous studies have shown that a single threshold for assigning acute lymphoblastic leukemia (ALL) patients to an MRD risk group failed to reflect the relapse risk of the different genetic subtypes.²² Similarly, in this study, we found that similar MRD levels had distinct prognostic implications for RUNX1::RUNX1T1 and CBFB::MYH11 subtypes. Specifically, a less than 3-log MRD reduction of correlated with an increased risk of relapse and decreased OS for patients with RUNX1::RUNX1T1, while it did not negatively affect the survival of CBFB::MYH11 patients. The ELN MRD guidelines emphasize the importance of tailored MRD monitoring strategies, recommending distinct time points for MRD assessment based on genetic subgroups.²¹ Our findings align with these recommendations, particularly in underscoring the need for genetic context when interpreting MRD results. However, the optimal timing and frequency of MRD assessments may differ between these subgroups, as reflected in the ELN guidelines. Future risk assessment algorithms for CBF-AML should incorporate both MRD levels and genetic background to more accurately identify patient relapse risk. Additionally, future studies should explore the integration of serial MRD monitoring into treatment protocols, ensuring that MRD dynamics are captured comprehensively to refine risk stratification and therapeutic strategies.

Although HSCT during CR1 has demonstrated success in reducing relapse risk and improving survival in AML patients with certain risk factors, it is not routinely recommended for CBF AML. However, for CBF-AML patients with an less than 3-log MRD reduction after two induction cycles, the ELN guidelines recommend HSCT due to their higher relapse risk.^21,23-25 Our study also confirmed that HSCT benefits CBFB::MYH11 AML patients with a <3-log MRD reduction.

However, for RUNX1::RUNX1T1 patients, the impact of HSCT was influenced by the induction regimen. Among patients treated with SD cytarabine, HSCT did provide additional benefits for those with a <3-log MRD reduction after two courses of chemotherapy, while for patients receiving ID regimens, the same MRD threshold could not reliably determine the necessity for transplantation. Our previous study also suggested that AML patients with flow cytometry-detected MRD positivity in different risk groups have different criteria for transplantation stratification.¹³ Therefore, we propose that different types of AML patients treated with varying regimens may require distinct MRD thresholds to guide transplantation decisions.

Mutations in kinase-related genes are the most common molecular genetic abnormalities in patients with CBF-AML.⁷ In the present study, we have corroborated previous findings that KIT mutations, particularly those in exon 17 (Online Supplementary Figures S13 and S14), correlate with poor prognosis and NRAS mutations with better outcomes in RUNX1::RUNX1T1 patients.^11,26-29 Additionally, we have identified TET2 mutations as a prognostic marker indicative of poor outcomes in patients with RUNX1::RUNX1T1. Previous literature has reported an association between TET2 mutations and advanced age.^30,31 However, our data may not provide evidence for older age in patients with TET2 mutations, likely due to limitations in age distribution and insufficient sample size. Further investigations involving larger, age-diverse cohorts are needed to validate our conclusions. Previous studies have found ZBTB7A mutations in 13 of 56 and four of 41 RUNX1::RUNX1T1 AML, respective-ly.^32,33 In this study, we reported ZBTB7A mutations in 14 of 114 RUNX1::RUNX1T1 AML cases, with a mutation frequency of approximately 12%. The presence or absence of this mutation did not appear to significantly affect the patient survival.

Our study has several limitations. First, it was a retrospective, single-center study. While this design enables consistent data collection and analysis, the findings should be generalized cautiously to broader populations or different clinical settings. Second, our findings were based on data from patients aged 60 years or younger who received intensive chemotherapy. Whether these results apply to older patients or those undergoing lower-intensity therapy remains to be investigated. Additionally, the unavailability of GO in our country precluded a comparison of ID induction regimens with those containing GO. Finally, the relatively small sample size of patients with targeted DNA sequencing limited the granularity of our mutational spectrum analysis.

Footnotes

• Received December 30, 2024
• Accepted April 22, 2025

Correspondence

^*MY, WW and GZ contributed equally as first authors

H. Wei weihui@ihcams.ac.cn

J. Wang wangjx@ihcams.ac.cn

Y. Wang wangying2@ihcams.ac.cn

Disclosures

No conflicts of interest to disclose.

Contributions

HW and JW participated in concept design. MY participated in data analysis, drafting and revising the manuscript. WW, YW and GZ collected and organized the clinical data. SQ, BL, and YM interpreted the results. All authors have read and approved the final manuscript.

Funding

The study was funded by the National Key Research and Development Program of China (2023YFC2508900), the National Natural Science Foundation of China (82370183), CAMS Innovation Fund for Medical Sciences (2023-I2M-2-007), Tian Jin Natural Science Foundation (23JCZXJC00310), Haihe Laboratory of Cell Ecosystem Innovation Fund (22HHXBSS00040), and Beijing Xisike Clinical Oncology Research Foundation (Y-SYBLD2022ZD-0031).