Dear editor,

CBF-AMLs, which account for approximately 12–15% of all AML cases, involve chromosomal translocations or inversions that target the transcription factors RUNX1 and CBFB [1]. CBF-AMLs is considered chemosensitive. However, some patients are not eligible. The venetoclax-based regimen has emerged as a promising treatment for frail AML patients [2]. The fusion products found in CBF-AMLs block myeloid differentiation but are not sufficient to induce leukemia alone [3]. CBF-AMLs frequently harbor tyrosine kinase pathway mutations, including KIT, FLT3, and NRAS/KRAS mutations. These mutations have proven to be necessary to confer a proliferation and survival advantage on transformed cells [4] and are reported to adversely impact outcomes in CBF-AMLs [3, 5]. However, to our knowledge, no prior reports have compared the effects of venetoclax and intensive chemotherapy on the prognosis of CBF-AML patients with kinase mutations.

In this study, data from 142 newly diagnosed CBF-AML patients were collected between December 2015 and December 2023. 70 patients received induction with idarubicin-based (IA) regimen, who were from clinical trial NCT02323022. 72 patients received induction with venetoclax-based (VA) regimen, who were from a real-world study (No.2022219).

Patients in the VA cohort received azacitidine at 75 mg/m2 for 7 days or decitabine at 30 mg/m2 for 5 days and venetoclax daily. The dose of venetoclax was 100 mg on Day 1 and increased stepwise over 3 days to reach the target dose of 400 mg on days 3–28. In the IA cohort, patients received IDA 10–12 mg/m2 for 3 days and Ara-C 100 mg/m2 for 7 days intravenously.

The primary objective of this study was to evaluate the efficacy and safety of VA and IA regimens. This included the complete response (CR) rate, the partial remission (PR) rate, the no response (NR) rate, measurable residual disease (MRD) and the adverse events. The absolute copy numbers of the fusion gene transcripts were normalized by log10. The secondary objectives included overall survival (OS), event-free survival (EFS), and relapse-free survival (RFS).

An overview of the study and the characteristics of all patients, divided by CBF type and treatment regimen, are summarized in Fig. S1 and Table S1. In the CBFB::MYH11 cohort, 39 patients (68.4%) received VA regimen, whereas 18 patients (31.6%) received IA regimen as induction treatment. Thirty-three patients (38.8%) and 52 patients (61.2%) in the RUNX1::RUNX1T1 cohort received VA and IA regimens, respectively. Briefly, in the two CBF-type cohorts, the clinical and laboratory characteristics were similar between these two groups. Except for FAB risk stratification in patients with RUNX1::RUNX1T1, more patients who received venetoclax were M1 or M5 (P = 0.048). The cytogenetic and molecular genetic characteristics are summarized in Table S2 and Fig. S1. KIT mutation (43.0%, n = 61) was the most frequently identified mutation, followed by NRAS (19.0%, n = 27), FLT3-ITD (13.4%, n = 19), and FLT3-TKD (12.0%, n = 17) mutations in CBF-AML patients. In our study, patients who achieve morphological CR again after recurrence but remain MRD positive or those whose fusion gene transcripts remain persistently positive after induction remission receive HSCT.

Among all CBFB::MYH11 patients, CR, PR, and NR rates were 94.9%, 5.1%, and 0%, respectively, in the VA group and 88.9%, 11.1%, and 0%, respectively, in the IA group (P = 0.184; Fig. 1A) after the first course of induction. In patients with KIT, FLT3-ITD, or NRAS mutations, CR rate was achieved in 96.6% of the patients in the VA group and 100.0% of the patients in the IA group (P = 0.001; Fig. 1B); in those without KIT, FLT3-ITD, or NRAS mutations, the CR incidence was 90.0% and 77.8%, respectively (P = 0.334; Fig. 1C). CR was achieved in all the patients (P = 0.001; Fig. 1A–C) after the second course of therapy. Among all RUNX1::RUNX1T1 patients, CR, PR and NR rates were 51.5%, 27.3%, and 21.2%, respectively, in the VA group and 82.7%, 15.4%, and 1.9%, respectively, in the IA group (P = 0.002; Fig. 1D) before the beginning of cycle 2. In patients with KIT, FLT3-ITD, or NRAS mutations, complete remission was achieved in 42.9% of the patients receiving the VA regimen and in 86.2% of the patients receiving the IA regimen (P = 0.004; Fig. 1E); in those without KIT, FLT3-ITD, or NRAS mutations, the CR incidence was 66.7% and 78.3%, respectively (P = 0.172; Fig. 1F).

Fig. 1: Response outcomes of the CBF-AML patients after induction therapy.
figure 1

A response outcomes of all patients with CBFB::MYH11. B, C response outcomes of CBFB::MYH11 cohort with or without FLT3-ITD/NRAS/KIT mutation. D response outcomes of all patients with RUNX1::RUNX1T1. E, F response outcomes of RUNX1::RUNX1T1 cohort with or without FLT3-ITD/NRAS/KIT mutation.

Full size image

Among all patients with CBFB::MYH11, median BM log10-transformed TLs were not significantly different between the VA group and the IA group (1.92 vs. 1.60; P = 0.065; Fig. 2A) after cycle 1 or after cycle 2 (1.29 vs. 1.00; P = 0.177; Fig. 2A). We next evaluated the impact of concurrent kinase mutations (KIT, FLT3-ITD, or NRAS) on fusion gene transcription level (TL) kinetics in BMs. There were no differences in TLs in the BM after two courses of induction (Fig. 2B), but the difference reached statistical significance in patients without kinase mutations after cycle 1 (2.04 vs. 0.00; P = 0.002; Fig. 2C). When analyzing log10-transformed RUNX1::RUNX1T1 TLs from the BM, we observed a lower level of TLs in the BM of the IA patients (4.07 vs. 4.48; P = 0.108; Fig. 2D) after cycle 1 and after cycle 2 (1.41 vs. 2.13; P = 0.019; Fig. 2D). Furthermore, kinase mutations (KIT, FLT3-ITD, or NRAS) were associated with low BM TLs (cycle 1: 4.15 vs. 4.57; P = 0.013; cycle 2: 1.41 vs. 4.68; P = 0.007; Fig. 2E). Among patients without kinase mutations, there were no differences in the number of TLs in the BM after induction (cycle 1: 2.98 vs. 3.71; P = 0.475; cycle 2: 1.83 vs. 1.30; P = 0.110; Fig. 2F).

Fig. 2: The chimeric transcript level of the CBF-AML patients after induction therapy.
figure 2

A The chimeric transcript level of all CBFB::MYH11 patients. B, C The chimeric transcript level of CBFB::MYH11 cohort with or without FLT3-ITD/NRAS/KIT mutation. D The chimeric transcript level of all patients with RUNX1::RUNX1T1. E, F The chimeric transcript level of RUNX1::RUNX1T1 cohort with or without FLT3-ITD/NRAS/KIT mutation.

Full size image

Common adverse events are summarized in Table S3. Among all the patients, the most frequently reported hematologic adverse events included thrombocytopenia, neutropenia, and anemia. Gastrointestinal adverse events were common and predominantly included nausea and vomiting.

In CBFB::MYH11-positive patients who underwent HSCT, the median OS was 16.4 months in the VA and and 25.8 months IA groups (P = 0.07, Fig. S2A). The median EFS was 14.0 months versus 20.0 months (P = 0.079, Fig. S2B). The median RFS was 11.6 months versus 18.5 months (P = 0.086, Fig. S2C). Among patients who did not undergo HSCT, the median OS was 8.7 months and 5.6 months in the VA and IA groups, respectively (P = 0.022, Fig. S2A). The median EFS was 9.1 months versus 5.6 months (P = 0.022, Fig. S2B). The median RFS was 7.8 months versus 36.0 months (P = 0.56, Fig. S2C). In RUNX1::RUNX1T1-positive patients who underwent HSCT, the median OS was 20.3 months and 20.1 months in the VA and IA groups, respectively (P = 0.38, Fig. S2D). The median EFS was 12.4 months versus 16.7 months (P = 0.51, Fig. S2E). The median RFS was 11.1 months versus 16.3 months (P = 0.57, Fig. S2F). Among patients who did not undergo HSCT, the median OS was 7.1 months versus 9.1 months in the VA and IA groups, respectively (P = 0.32, Fig. S2D). The median EFS was 6.8 months versus 9.1 months (P = 0.36, Fig. S2E). The median RFS was 5.9 months versus 9.8 months (P = 0.43, Fig. S2F).

Recent genomic studies highlighted considerable heterogeneity between the two subtypes [6]. In our study, the genetic discrepancy between the two CBF-AML entities became apparent when all recurrently mutated genes were segregated into functional subgroups categorized. The only major mutational overlap occurred in RTK/RAS signaling genes, which was mostly due to mutations in NRAS, KIT, and FLT3.

Previous work revealed that the fusion transcripts CBFB::MYH11 and RUNX1::RUNX1T1 were not sufficient for leukemia development and that additional mutations, such as KIT, FLT3-ITD, and NRAS, were required for its onset during leukemogenesis, promoting proliferation [3, 7,8,9,10]. However, several groups have reported that the prognostic impact of these driver mutations differs between patients within each subset of CBF-AML [11,12,13]. Similar to previous reports, the results of the present analysis support the notion that kinase mutations are associated with poor results in patients with RUNX1::RUNX1T1. Furthermore, our data clearly revealed that, compared with the IA regimen, the VA regimen was associated with a worse response and higher fusion gene TLs in RUNX1::RUNX1T1-positive AML patients with kinase mutations after both courses of induction treatment. Moreover, we demonstrated that there was no difference in outcomes among patients without kinase mutations. Among CBFB::MYH11 AML patients, regardless of the mutation or risk factors, there was no difference in the induction response or MRD between the two regimens. In line with those of previous studies, a higher frequency of hematologic adverse events in the IA group [14, 15].

In conclusion, VA treatment had shown same efficacy and greater safety than IA in CBF-AML, expect for RUNX1::RUNX1T1-positive patients with kinase mutations. Because of the limited sample size and the fact that many nonresponders switched to other regimens after the first cycle of VA, the results of this study warrant an investigation in a prospective randomized controlled study.