This is a single center, single arm, phase 2 trial at Zhongda Hospital (Nanjing, China). This trial was registered on ClinicalTrials.gov, number NCT04248595.

A total of 20 AML patients (aged 18-70 years) diagnosed by the 2016 World Health Organization criteria (WHO) (24) were enrolled between Jan 2020 and Sep 2022. All of the enrolled patients were ND, untreated de novo AML (no history of the myelodysplastic syndrome (MDS), myeloproliferative neoplasms (MPN), chronic myelomonocytic leukemia (CMML), or exposure to potentially leukemogenic agents). The risk categories of the patients were based on the European LeukemiaNet 2022 (ELN 2022) risk category Guidelines for AML (25). The study design is shown in Figure 1A .

This study was approved by the Institutional Review Board (approval code: 2019ZDSYLL211-P01) and all patients provided written informed consent before enrollment according to institutional policies and the declaration of Helsinki.

Induction therapy in a 28-day cycle consisted of Aza (75mg/m²/d on days 1-5 subcutaneous) administered in combination with HIA regimen [HHT 2mg/m²/d on days 4-8 intravenous over 3 hours, idarubicin 6mg/m²/d on days 4-6 intravenous, and cytarabine 100mg/m²/d on days 4-10 intravenous] for patients age<60 or age≥60 eligible for IC ( Figure 1A ). Patients who did not achieve a CR or CR with incomplete blood count recovery (CRi) following the first cycle could receive a second cycle at the same doses. The patients who did not reach the CR/CRi after the second cycle were withdrawn from the study. Post-remission therapy for patients who received CR/CRi, high-dose cytarabine was given as consolidation therapy followed by hematopoietic stem cell transplantation for those candidates ( Figure 1B ).

In this study, to prevent infection of patients with granulocytopenia, quinolones antibiotics, voriconazole (or echinocandins) were considered during the induction until the recovery of neutrophils, and laminar flow air cleaner beds were utilized to reduce opportunistic infection.

Response to therapy was monitored by analysis of blood and bone marrow (BM) aspirates. Response assessment was done at the end of cycles 1 and 2, after every 2 cycles of consolidation and every 3 months during maintenance to confirm ongoing response. Responses were categorized based on the revised International Working Group criteria for AML (26).

Minimal residual disease (MRD) assessment by multicolor flow cytometry (MFC) was done on pretreatment BM and all subsequent BM examinations as previously described (27). BM samples were obtained at diagnosis and evaluated for the presence of cytogenetic and molecular aberrations. Real-time quantitative PCR (RT-qPCR) was used to detect the gene rearrangements. For documentation of mutations, the entire coding sequences of 58 leukemia driver genes ( Table S1 ) known to be frequently mutated in myeloid malignancies were sequenced with targeted leukemia exome-seq panel.

Adverse events (AEs) and laboratory values, graded according to the Common Terminology Criteria for AEs version 4.0, were evaluated at least once every cycle during induction and consolidation and then at least every 3 months during the maintenance.

In this trial, we applied minimax Simon’s two-stage design method (28) to determine the sample size. A composite CR (CR or CRi) rate ≤ 65.0% (p0) would be unacceptable (null hypothesis). While a composite CR rate of higher than 90.0% (p1) would support the ongoing trial for further exploration. Meanwhile, we used an estimation power of 80.0%, with a significance level of 5.00% to test the hypothesis.

15 participants were enrolled in the 1st stage, with additional 3 participants in the 2nd stage. This trial would be discontinued if 12 or fewer achieved composite CR in the 1^st stage. The enrollment procedure of the 2nd stage would be held until the interim analysis result of the 1st stage. If this trial continued, the activity of the combination therapy would be considered a null hypothesis if less than 14 out of 18 enrolled cases achieved composite CR. Finally, considering a 10% dropout rate, the estimated sample size was 20 AML patients.

The primary endpoint was the composite CR (CR/CRi). Criteria of CR were as follows: (1) no signs of leukemia after treatment, (2) fewer than 5% blasts in the BM, (3) absolute neutrophil count (ANC) >1000/μL and platelet count > 100,000/μL, (4) absence of the extramedullary tumor. CRi was defined as fulfilling all CR criteria except for ANC <1000/μL or platelet count <100,000/μL in peripheral blood. Secondary endpoints were overall survival (OS) defined as the time from study entry to death from any cause, relapse-free survival (RFS) defined as the time from achieving a CR until disease recurrence or death, AEs (including hematological and non-hematological AEs defined as any unfavorable and unintended signs including an abnormal laboratory finding, symptom, or disease). Treatment failure was defined as not achieving CR or CRi after two cycles of induction therapy.

Prespecified correlative assessments included targeted gene panel sequencing to assess associations between somatic mutation patterns and therapeutic responses as well as disease progression.

Eligible patients were aged 18-70 years and ND, previously untreated de novo AML according to the criteria of WHO (24). The including criteria were as follows: (1) Eastern Cooperative Oncology Group performance status (ECOG) ≤2 (5-point scale), (2) Normal left ventricular ejection fraction ≥50% (echocardiography), (3) Adequate hepatic function (alanine aminotransferase and aspartate aminotransferase concentration < 2 times of upper limit of normal, serum bilirubin concentration ≤35 μmol/L, (4) Adequate renal function (serum creatinine ≤150 μmol/L), (5) Able to understand and provide the written consent. Exclusion criteria were as follows: (1) Patients diagnosed with acute promyelocytic leukemia (M3), (2) Patients with a history of the antecedent myeloid disease, including MDS, MPN, or CMML, (3) Patients were positive for HIV (human immunodeficiency virus), (4) A history of chronic stable angina or congestive heart failure with an ejection fraction < 50%, or COPD (chronic obstructive pulmonary disease) with FEV (forced expiratory volume) less than 65% in 1 second, (5) Uncontrollable infection (bacterial, fungal or viral) or respiratory disease (requiring continuous supplementary oxygen), (6) Mentally unhealthy or unable to sign the informed consent, (7) lactating and pregnant women (women of childbearing age would receive the test of β-subunit of human chorionic gonadotropin), (8) Other medical or psychological diseases which investigators believed were not suitable to enter the trial.

A leukemia targeted-exome-seq panel including 58 leukemia driver genes was used for screening the gene mutations in 20 enrolled patients by next-generation sequencing (NGS) before and after the induction therapy (29, 30). The target genes in the panel are listed in Table S1 .

Agilent SureSelect Human All Exon V4+UTRs (Agilent) was used for the coding exons plus UTRs of target genes. Probes for each exon of each target gene are designed by the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/). The targeted exome-seq method is performed as reported (31). Briefly, the genomic DNA was isolated from BM samples with the genomic DNA isolation kit (Qiagen, Hilden, Germany). All DNA samples were sheared with a Covaris E220 instrument generating approximately 260 bp DNA fragments. The fragmented DNA was processed into Illumina-compatible sequencing libraries using Kapa Hyper Prep Kit (Illumina, San Diego, CA, USA). Each library was uniquely barcoded and captured by the leukemia panel probes, followed by PCR amplification and sequencing on a HiSeq 2500 (Illumina) with 2x100 bp reads. The sequencing reads were aligned to the human genome by following Broad Institute’s GATK best-practice pipeline to call germline short variants (SNPs and Indels). Called variants were annotated using ANNOVAR (version 2.3). Exonic variants with exonic, nonsynonymous, stop-gain, or stop-loss, novel SNPs, and with predicted deleterious/damaging functions were manually surveyed by the Integrative Genomics Viewer (IGV) to confirm.

The association of gene mutations with clinical response, relapse, and risk status was analyzed with R 4.0 software and depicted as a waterfall figure. The association of the gene mutations with OS and RFS was also evaluated by the Kaplan-Meier method.

MRD assessment by MFC was done on pretreatment BM and all subsequent BM examinations with an assay sensitivity of 0.01% (32). A wide antibody panel of molecular antigens (Beckman, USA) used for MRD detection of AML were as follows: CD34, CD117, CD13, CD33, CD7, CD56, CD10, CD19, CD38, HLA-DR, and CD45. After incubating the BM samples with monoclonal antibodies for 15 minutes, erythrocytes were lysed using lysing solution (QIAGEN Science, USA). Afterward, to remove excess antibodies and lysed red blood cells, the samples were washed twice with 2 mL PBS (phosphate-buffered saline) buffer (pH 7.4) and centrifuged (600g×5 minutes, room temperature). The samples were then measured using the NAVIOS system (Beckman Coulter) and analyzed for at least 50, 000 events. All samples were analyzed on the Kaluza analyzing platform (Beckman, USA). The MRD detected by MFC was defined as at least 20 clustered cells showing myeloid scattering properties, leukemia-associated immunophenotype (LAIP) characteristics, and a minimum of 1×10⁶ white blood cells (WBC) were required for MRD monitoring to achieve theoretically maximum sensitivity of 2×10^-5. After checking cell viability and debris through FSC/SSC scatter plot, the main gating strategy for detecting MRD was as follows: 1. CD45/SSC plot was used to show the whole WBC, 2. CD34+, CD117+, or CD133+ fractions were used to show progenitor/primitive cells, 3. CD45dim/SSC low plot was used to detect immature cells (or CD45high/CD34 negative plot if cells are more mature), 4. within the above fractions, the percentage of aberrancies was defined by the plot of myeloid markers (CD33 and CD13) in combination with lineage markers (CD7, CD56). The negative status of MRD was defined as MRD less than 0.1%, while the positive status of MRD was defined as 0.1% or higher. Two investigators were joined to evaluate all the samples to avoid bias.

A leukemia fusion gene panel including 47 fusion genes was used for screening all 20 enrolled patients at the time of diagnosis. The fusion genes in the panel are listed in Table S2 .

BM samples were obtained at diagnosis and evaluated for the presence of gene rearrangement. The isolation of total RNA, cDNA synthesis, and RT-qPCR was used to detect the gene rearrangements, according to the manufacturer’s (Zeesan Biotech, China) protocol. The RT-qPCR was performed on the ABI7500 platform (thermos scientific, USA) with specific primers for the fusion genes. Each procedure of PCR included a standard curve (a serial dilution of plasmids from 10⁶ to 10²), negative controls, and patients’ samples. Negative controls and patients’ samples were in duplicate and standard curves in triplicate. The sensitivity of RT-qPCR was 10^-5 at the cellular level. Patients with specific positive fusion genes were further monitored after the induction therapy by the RT-qPCR method, and the absolute copy number of fusion gene transcripts was normalized to ABL1, expressed per 10⁵ copies of ABL1.

Lower and upper limits of the 95% confidence interval for a proportion were calculated by the Wilson method (continuity correction). Survival was estimated by the Kaplan-Meier method, which was done by STATA 14.0 and GraphPad 8.0 software. A two-sided α level of 0.05 was considered statistically significant.

Between Jan 2020 and Sep 2022, 26 patients were screened for eligibility, and 20 previously untreated de novo AML patients were enrolled (9 were male, 11 were female). The median age was 50 years (range 18-70) (3 of 20 patients ≥60 years, 17 of 20 patients <60 years). The enrolled patients were classified into favorable-risk (9/20), intermediate-risk (8/20), and poor-risk(3/20)according to the ELN 2022 risk category Guidelines for AML. The most frequently mutated genes were FLT3 (ITD/TKD) (20.0%, 4/20), DNMT3A (20.0%, 4/20), NRAS (15.0%, 3/20), and KRAS (15.0%, 3/20). The baseline characteristics and demographic of the enrolled patients are presented in Table 1 .

CR/CRi was achieved in 95.0% (19/20; [95%CI, 73.1% to 99.7%]) of the enrolled patients ( Figure 2A ). 94.7% (18/19) reached CR/CRi after the 1st cycle of the induction therapy and the median time to the first CR/CRi was 25.0d (IQR, 21.0 to 33.0). At the date cutoff (Oct 2, 2022), the median follow-up was 733 days (range, 30 to 1059) for all enrolled patients, and no patient was lost to the follow-up. The median OS and RFS were not reached and the estimated 2-year OS and RFS were 87.5% (95%CI, 58.6% to 96.7%) and 87.1% (95%CI, 57.3% to 96.6%), respectively ( Figures 2B, C ).

Based on ELN2022 risk classification, CR/CRi rate was achieved in 100% (17/17) of the favorable and intermediate risk-group, and 66.7% (2/3) in the poor risk-group. No CR/CRi rate difference was observed between the favorable and intermediate versus poor-risk groups (P=0.15). Besides, no survival difference was observed among the different risk groups in both OS (favorable vs intermediate, P=0.2801; intermediate vs poor, P=0.666; favorable vs poor, P=0.1266) and RFS (favorable vs intermediate, P=0.2482; intermediate vs poor, P=0.4581; favorable vs poor, P=0.0614) ( Figure 2D, E ).

We detected 36 mutants in 19 of 58 leukemia driver genes in all enrolled patients at the timepoint of enrollment. The median mutation number/patient was 2.0 (range, 0-5), and 18 of 20 (90.0%) patients had more than one mutation (≥1) ( Figure 3A ). More than half of the patients (11/20) harbored the mutated genes involved in signal transducers (FLT3, KRAS, NRAS, KIT, PTPN11, CSF3R; 11/20, 55.0%), and all these patients achieved CR/CRi (11/11, 100%) upon the novel combined therapy ( Figure 3A ). Also, the 11 patients who harbored mutations involved in the signal transducers acquired an improved trend in both OS (HR=0.126, 95%CI, 0.00786 to 2.021; P=0.1435) ( Figure 3B ) and RFS (HR=0.0993, 95%CI, 0.00606 to 1.629; P=0.1056) ( Figure 3C ), although no significance was observed compared to that without the mutants. It is worth noticing that a high mutant’s clearance with VAF (variant allele frequency) was dramatically reduced (to undetectable) ( Figure 3D ) after the induction therapy among all of the 11 cases. These data indicate that the patients with the gene mutation of the signal transducers had a rapid and high clinical response to the Aza+HIA regimen.

We also screened the 47 common leukemia fusion genes in the trial and observed the efficacy in the patients with leukemia fusion genes. We identified 3 patients harboring the MLL-AF9 fusion gene and 3 patients having CBF-β -MYH11 fusion gene. All 3 patients with the MLL-AF9 fusion gene reached complete molecular remission (<0.1%, undetectable) after 1 cycle of induction. 2 of 3 patients with the CBF-β -MYH11 fusion gene reached complete molecular remission (<0.1%, undetectable), and the expression level of the CBF-β -MYH11 fusion gene in 1 patient decreased over 1000-fold at the molecular level (512.16% to 0.28%) after 1 cycle of induction. These results indicated that the patients with the fusion genes could also achieve clinical and/or molecular remission upon the Aza+HIA induction treatment.

The MRD negative was 89.5% (17/19; 95% CI, 65.5% to 98.2%) in patients who reached CR/CRi, and 88.9% (16/18; 95% CI, 63.9% to 98.1%) in patients who reached CR/CRi after the 1^st cycle of induction therapy ( Figures 4A, B ). Also, MRD was negative in 88.9% of patients with favorable-risk (8/9; 95%CI, 50.7% to 99.4%), 100% (8/8; 95%CI, 59.8% to 99.9%) with intermediate-risk, and 50.0% (1/2; 95%CI, 2.67% to 97.3%) with poor-risk ( Figure 4C ). Moreover, MRD negative was achieved in one patient after one cycle of consolidation therapy. These data indicated the novel regimen could efficiently eradicate the MRD to reach the deep and rapid molecular response in de novo AML.

Common non-hematological AEs are summarized in Table 2 , infection (grade 1-4) was the most common non-hematological AE, presenting 75.0% (15/20) in the whole cohort, and infection of grade 3 or higher occurred in 60.0% of patients (12/20). Other AEs of grade 3 or higher that occurred in the whole cohort included fatigue (1/20, 5.0%), decreased appetite (1/20, 5.0%), and hepatic dysfunction (1/20, 5.0%). No grade 5 AEs or tumor lysis event was observed, and no cardiac AE was observed.

As for the hematological toxicity, the most common grade 3-4 hematological AEs were neutropenia (100%, 20/20) and thrombocytopenia (100%, 20/20). Among 19 patients who reached CR/CRi, the median duration of neutropenia (ANC ≤500/μL) was 12d (IQR, 11 to 15), and the median duration of thrombocytopenia (absolute platelet count ≤50000/μL) was 14d (IQR, 13 to 16).

It is noticed that no patients discontinued the induction treatment or required dose reduction due to hematological or non-hematological AEs. In addition, no patient died within 8 weeks of the treatment.