Venetoclax in Combination with Cytarabine with or without Idarubicin or Azacitidine in Children, Adolescents, and Young Adults with Relapsed or Refractory Acute Myeloid Leukaemia: a Multi-Center Phase I Expansion Study

Abstract

Background:

The prognosis for paediatric patients with relapsed or refractory acute myeloid leukaemia remains poor. While encouraging, published paediatric data on venetoclax combined with intensive chemotherapy is limited. This study aimed to refine efficacy and toxicity estimates from the previously published dose escalation phase of this study.

Methods:

In this multicentre trial (NCT03194932), paediatric patients with relapsed/refractory AML received venetoclax 360mg/m² (max 600mg) daily for 27 days after a 50% dose reduction on day 1 and cytarabine 1g/m²/dose q12h for 8 doses beginning day 8, with idarubicin or azacitidine included in predefined treatment cohorts. Here, we report results for all patients treated at the recommended phase 2 dose (RP2D), including patients treated in a phase 1 expansion cohort and patients in a cohort combining venetoclax with high-dose cytarabine and azacitidine added by protocol amendment. This final report describes the key secondary endpoints and intention-to-treat response rates at the RP2D.

Findings:

From 07/1/2017 to 7/22/2022, 44 patients (21 female, 22 White, 6 Black, 14 Hispanic) were treated at the RP2D. The median follow-up is 5.3 years. After 1 cycle, 25 (56.8%, 95% confidence interval 41.1-71.7%) patients achieved a complete response with or without haematological recovery; 19 were minimal residual disease negative. The overall response rate was 65.9% (50-79.6%). Common grade 3/4 adverse events included febrile neutropenia (23/44, 52.3%), colitis (4/44, 9.1%), and infections (11/44, 25%). Toxicity from venetoclax combined with high-dose cytarabine and azacitidine was acceptable.

Interpretation:

Venetoclax with high-dose cytarabine is active with acceptable safety in paediatric patients with relapsed/ refractory AML. Inter-group differences between cohorts and limited cohort sizes preclude definitive inter-group comparison. These findings support ongoing research of venetoclax/high-dose cytarabine regimens in this population, including the randomized paediatric phase 3 ITCC-101/APAL2020D trial (NCT05183035).

Funding:

US NIH (CA021765), ALSAC, AbbVie, Gateway for Cancer Research.

Introduction

Children, adolescents, and young adult patients with acute myeloid leukaemia (AML) have survival rates that have plateaued at approximately 70%, with relapse rates between 20-40%.^1-6 Among relapsed patients, long-term survival remains poor, with 5-year overall survival rates of only 35%.⁷ The genomic heterogeneity of relapsed paediatric AML underscores the urgent need for broadly applicable therapies to improve outcomes.⁸ Venetoclax, a selective BCL-2 inhibitor with demonstrated activity across diverse adult AML subtypes and regulatory approval in multiple adult AML settings, represents a rational therapeutic strategy.

The over-expression of anti-apoptotic BCL-2 family proteins is a recurrent mechanism of chemotherapeutic resistance in AML and other malignancies.^9,10 Preclinical work shows that BCL-2 inhibitors have activity against AML cell lines and patient-derived samples.^9,10 Furthermore, venetoclax combined with low-dose cytarabine or hypomethylating agents were superior to chemotherapy alone in unfit patients with AML, resulting in FDA approval for these combinations.^11,12 Paediatric phase I data has shown signals of efficacy.^13,14 Data from the first 38 patients treated in the phase I study of venetoclax in combination with cytarabine with or without idarubicin in children with relapsed/refractory AML showed that venetoclax could be safely combined with intensive chemotherapy in this population, and defined the recommended phase 2 dose (RP2D) now being evaluated in an ongoing phase III study.¹⁵ However, studies evaluating the combination of venetoclax with high-dose cytarabine chemotherapy regimens typical of paediatric relapse therapy remain scarce. Here, we present the outcomes of all patients treated with the RP2D of this completed phase I dose escalation and expansion study of venetoclax in paediatric patients with relapsed/refractory AML, providing refined estimates of toxicity and response and additional data on predictors of response.

Methods

Study design and participants

The VENAML phase 1 study included two phases and evaluated venetoclax-based combination regimens incorporating a high-dose cytarabine backbone, with additional agents introduced in predefined cohorts. Following completion of the dose escalation phase and determination of the recommended phase 2 doses (RP2D) previously reported,¹³ the dose expansion phase analysed outcomes by treatment cohort to further evaluate activity across regimens with differing therapeutic compositions.¹³ Here we report results from all patients enrolled on the RP2D including two expansion cohorts: cohort A received venetoclax with cytarabine and cohort B received venetoclax with cytarabine and idarubicin. The study was later amended, with institutional review board (IRB) approval, to allow patients to receive venetoclax with cytarabine and azacitidine (cohort C).

Patients with relapsed or refractory AML or acute leukaemia of ambiguous lineage (acute undifferentiated leukaemia or mixed phenotypic acute leukaemia) between 2 and ≤24 years of age were eligible. Refractory disease was defined as persistent leukaemia after at least two courses of induction chemotherapy. Relapsed disease was defined as the re-emergence of leukaemia after the achievement of remission. Eligible patients had ≥5% blasts in the bone marrow by morphology or ≥1% blasts by flow cytometry. If an adequate bone marrow sample was unobtainable, unequivocal evidence of leukaemia with ≥5% blasts in the peripheral blood was required. Standard organ function, Lansky/Karnofsky score ≥50, and prior therapy washouts were required. Patients with Down syndrome, acute promyelocytic leukaemia, juvenile myelomonocytic leukaemia, bone marrow failure syndromes, uncontrolled infections, or gastrointestinal conditions that would impair absorption of venetoclax were ineligible (see full inclusion and exclusion criteria in Supplement pages 1-2). A patient advocate reviewed the trial as part of scientific and IRB approvals. Patient age, sex, race, and ethnicity were provided by self-report.

Informed consent, and assent as appropriate, were obtained from all patients or their legal guardians in accordance with the Declaration of Helsinki. The institutional review boards of all three participating sites (see Supplement page 1) approved the protocol. This completed trial was registered at ClinicalTrials.gov (NCT03194932, https://clinicaltrials.gov/study/NCT03194932).

Procedures

All patients in cohorts A and B received venetoclax for 28 days administered at 180 mg/m² (max 300 mg) on day 1 and 360 mg/m² (adult equivalent dose 600 mg) once daily on days 2-28 with high-dose cytarabine given intravenously (1,000 mg/m² per dose every 12 hours for 8 doses) on days 8-11 with cohort specific differences. Cohort A received only venetoclax and cytarabine as described above; cohort B received these agents as well as idarubicin 12 mg/m² intravenously on day 8 along with 600 mg/m² of intravenous dexrazoxane and was restricted to patients who had previously received less than 300 mg/m² of doxorubicin equivalents. The study was later amended, with institutional review board (IRB) approval (10/13/2020), to allow patients to receive 75 mg/m² of azacitidine on Days 1-7 alongside 21 days of venetoclax with high-dose cytarabine on days 8-11 (cohort C). After cohort C was deemed safe, up to 6 additional patients could be treated to explore activity. Dose limiting toxicity definitions matched those in other cohorts in phase 1 with continuous toxicity monitoring and are described further below.¹³ Patients could receive up to four 28-day cycles or could proceed to alternative therapy including hematopoietic stem cell transplantation (HSCT) after any cycle.

Patients received intrathecal therapy during screening and those with evidence of central nervous system leukaemia received weekly intrathecal therapy until leukaemic blasts were cleared from the cerebrospinal fluid. Prophylactic antibacterial and antifungal therapy were given according to institutional guidelines, typically including levofloxacin and micafungin, with vancomycin added after cytarabine. Because venetoclax is predominantly metabolized by CYP3A4, the use of azole antifungals for prophylaxis was not permitted during venetoclax administration. Patients who required voriconazole or posaconazole for treatment of invasive fungal infections received 25% of the planned venetoclax dose (45 mg/m² on day 1 followed by 90 mg/m² for the remainder of treatment). Patients were evaluated with physical exam, blood counts and differential, and blood chemistries at least weekly.

Response was determined by bone marrow evaluation between days 35 and 49 in patients who completed the combination therapy. Patients with hypocellular marrow at first evaluation underwent a repeat marrow evaluation at the time of haematological recovery but not later than day 49. The final bone marrow evaluation was used to classify response. MRD for response determination was assessed by multiparameter flow cytometry per protocol and was performed centrally at St. Jude Children’s Research Hospital or at Hematologics, Inc., when applicable. For patients whose immunophenotype did not allow for evaluation of MRD, response was determined by bone marrow morphology.

Per protocol, patients with evidence of disease progression during the venetoclax monotherapy window were permitted to initiate combination chemotherapy early. Response to single agent venetoclax was assessed by comparing absolute peripheral blood leukaemic blast count on days 1 and 8. For patients without evaluable peripheral blood MRD, an optional day 8 bone marrow evaluation was used to compare MRD in bone marrow before therapy and on day 8. Absolute blast burden at each timepoint was calculated by multiplying the peripheral white blood cell count by the MRD measured at that timepoint. A window response was defined as the relative change in absolute blast burden between day 1 (pre-treatment) and day 8 (the end venetoclax monotherapy). A good window response was defined as a ≥50% reduction in blasts between days 1 and 8,¹³ while a poor window response was defined as <50% reduction. This threshold was based on the cutoff established within the dose escalation cohort.

BH3 profiling was performed on cryopreserved blood or bone marrow cells stained with leukaemia-specific antibodies and permeabilized. BAD, HRK, and NOXA peptides were tested for induction of cytochrome c loss from the mitochondria of leukaemic blasts that were made permeable to digitoxin. Cytochrome c release from the mitochondrial outer membrane as measured by multicolour flow-cytometry, identified dependence on BCL-2, BCL-xL, and MCL-1 in the presence of relevant peptides (see also Supplementary methods, pages 3-4). Patient reported quality of life measures were obtained from participants and/or proxies using Patient Reported Outcomes Measurement Information System (PROMIS) while hopefulness and optimism were measured using the Herth Hope Index as previously described.¹³

Grade 2 or higher non-haematological toxicities as defined by the Common Terminology Criteria for Adverse Events version 4.03 were recorded in the research database. Any grade 3 or higher non-haematological event attributable to venetoclax was considered a dose-limiting toxicity, except for: grade 3 or 4 infection, grade 3 or 4 nausea, vomiting, or diarrhoea manageable with supportive care, or grade 3 or 4 electrolyte disturbance resolving to grade 2 or less within 7 days. Failure to recover an absolute neutrophil count ≥300 per μL or platelet count ≥30,000 per μL by day 50 was also deemed a dose-limiting toxicity, unless attributable to underlying leukaemia or infection. Invasive fungal infection was categorized according to European Organization for Research and Treatment of Cancer (EORTC) consensus criteria, with an additional category of suspected invasive fungal infection, comprising diagnosed invasive fungal infection with imaging findings consistent with the condition but not meeting EORTC criteria.¹⁶

Outcomes

The primary endpoint of the phase I component of the study was the RP2D for each combination, with the doses for cohort A (phase I dose level 3) and cohort B (phase I dose level 4) already published and cohort C presented here. Secondary endpoints included the rates of complete response with or without haematological recovery at the RP2D and are the focus of this report. Complete response, defined as bone marrow minimal residual disease (MRD) <5% by flow cytometry, was divided into those with haematological recovery (absolute neutrophil count 500 per mL³ and platelets 75,000 per mL³, CR), and with incomplete haematological recovery (CRi). Partial response (PR) was defined as MRD 5-25% with a ≥50% reduction in leukaemic blasts with or without haematological recovery. MRD negative CR/CRi was defined as marrow disease undetectable or <0.1% by flow cytometry based on laboratory established standards.¹³ Overall response rate (ORR) included CR, CRi, and PR.^13,14 All responses were centrally reviewed in the Department of Pathology at St. Jude Children’s Research Hospital, Memphis, TN. Exploratory endpoints included biomarker studies, including leukaemia cell genomic analyses, BH3 profiling, and their association with response; quality of life assessments; and, previously reported,^13,17 the pharmacokinetics of venetoclax.

Statistical Analysis

A 12-evaluable patient cohort expansion size was selected to obtain additional clinical experience with and preliminary data on the regimen without extensively prolonging the study. It was not chosen to achieve a specified level of statistical power for hypothesis testing or precision for statistical estimates. The rolling-6 design was used to determine safety in cohort C. This report utilizes an intention to treat analysis evaluating any eligible patient regardless of protocol therapy received and analyses both all patients treated at the RP2D and in each discrete cohort. Descriptive statistics were computed for patient baseline characteristics with mean (SD) and median (interquartile range [IQR]) for continuous variables and frequency (%) for categorical variables. Binomial confidence intervals were reported for response and toxicity rates. A post-hoc analysis of overall survival from therapy initiation to death or last follow-up used the Kaplan-Meier method. The reverse Kaplan-Meier method was used to estimate quantiles of follow-up times.¹⁸ All analyses were done with R (R Foundation for Statistical Computing, Vienna, Austria), Windows version 4.4.0. Post-hoc analyses were performed to evaluate the association between venetoclax window response and subsequent response to combination therapy, describe toxicities among all RP2D treated patients, describe response in patients according to genomics and prior therapy, and describe survival. No formal statistical comparisons were planned for descriptive subgroup analyses. Analyses are based on data available on February 20, 2026.

Role of funding source

Funders had no role in study design, data collection, data analysis, data interpretation, or the writing of the report. Pharmacokinetic samples from this study were used for paediatric dose determination by AbbVie and these data have been previously reported.¹⁷ Full data access was available to KPD, SBP, HW, and SEK.

Results

Between July 1, 2017, and July 22, 2022, 61 patients were enrolled on trial across three sites: Stanford University Medical Center, University of North Carolina, and St. Jude Children’s Research Hospital. As shown in Figure 1 (CONSORT diagram), the results of 44 of those patients are included in this report with median follow-up for all patients of 5.3 (IQR 4.2-6.3) years. The first 38 patients treated, including 30 during the dose-escalation phase of the study, have been previously published including their characteristics and per-protocol outcomes. Of these, 22 treated at the RP2D are included in this manuscript.¹³ Forty-six patients with a median age of 12.2 (IQR 8.1-16.3) years were enrolled on either DL3 or DL4 during dose escalation or dose expansion cohorts A, B, or C. Two patients were determined to be ineligible after enrolment but before therapy initiation and did not receive study therapy: one enrolled in cohort A who was subsequently found to have a fusion characteristic of atypical acute promyelocytic leukaemia and one enrolled in cohort C who was unable to swallow venetoclax tablets due to age. Thus, 44 patients were treated at the RP2D: 22 received cohort A therapy (venetoclax plus high-dose cytarabine), 15 received cohort B therapy (venetoclax plus high-dose cytarabine with idarubicin), and 7 received cohort C therapy (venetoclax plus high-dose cytarabine with azacitidine) and are analysed below (Table 1).

Figure 1.

Consort Diagram

Table 1:

Baseline Characteristics of Patients Treated at the Recommended Phase 2 Dose

Demographics	Cohort A (N=22)	Cohort B (N=15)	Cohort C (N=7)	Overall (N=44)
Age (years)
Mean (SD)	10.8 (4.4)	10.1 (4.0)	18.3 (1.7)	12.1 (4.8)
Median [IQR]	11 [6.4, 14.3]	9.8 [6.5, 13.6]	18.2 [16.8, 20.3]	12.2 [8.1, 16.3]
Sex
Female	10 (45.5%)	8 (53.3%)	3 (42.9%)	21 (47.7%)
Male	12 (54.5%)	7 (46.7%)	4 (57.1%)	23 (52.3%)
Race
Asian	0 (0%)	0 (0%)	1 (14.3%)	1 (2.3%)
Black	2 (9.1%)	2 (13.3%)	2 (28.6%)	6 (13.6%)
Multiple Race (NOS)	0 (0%)	1 (6.7%)	0 (0%)	1 (2.3%)
Hispanic White	7 (31.8%)	7 (46.7%)	0 (0%)	14 (31.8%)
Non-Hispanic White	13 (59.1%)	5 (33.3%)	4 (57.1%)	22 (50.0%)
Prior HSCT
Yes	9 (40.9%)	5 (33.3%)	2 (28.6%)	16 (36.4%)
No	13 (59.1%)	10 (66.7%)	4 (71.4%)	28 (63.6%)
Genetics *
KMT2A rearranged	7 (31.8%)	4 (26.7%)	0 (0%)	11 (25.0%)
RUNX1::RUNX1T1	2 (9.1%)	2 (13.3%)	0 (0%)	4 (9.1%)
NPM1 mutations	3 (13.6%)	0 (0%)	1 (14.3%)	4 (9.1%)
FLT3 ITD	1 (4.5%)	1 (6.7%)	2 (28.6%)	4 (9.1%)
NUP98 rearranged	0 (0%)	3 (23.1%)	0 (0%)	3 (6.8%)
CEBPA mutated	2 (9.1%)	0 (0%)	1 (14.3%)	3 (6.8%)
CBFß::MYH11	2 (9.1%)	0 (0%)	0 (0%)	2 (4.5%)
FLT3 point mutation	1 (4.5%)	1 (6.7%)	0 (0%)	2 (4.5%)
PICALM::MLLT10	1 (4.5%)	1 (6.7%)	0 (0%)	2 (4.5%)
UBTF TD	2 (9.1%)	0 (0%)	0 (0%)	2 (4.5%)
MECOM rearranged	0 (0%)	1 (6.7%)	0 (0%)	1 (2.3%)
Disease Status at Entry
Relapsed	13 (59.1%)	10 (66.7%)	4 (57.1%)	27 (61.4%)
Primary Refractory	0 (0%)	3 (20.0%)	1 (14.3%)	4 (9.1%)
Relapsed Refractory	9 (40.9%)	2 (13.3%)	2 (28.6%)	13 (29.5%)

Abbreviations: DL3, dose level 3; DL4, dose level 4; SD, standard deviation; Min, minimum; Max, maximum; NOS, not otherwise specified; HSCT, hematopoietic stem cell transplant; ITD, internal tandem duplication; TD, tandem duplication

^*Listed genetic features are mutually exclusive except for FLT3 internal tandem duplications and point mutations.

Three patients initiated study therapy but did not complete combination treatment due to progressive disease (two patients; cohorts A and C) or unacceptable toxicity (one patient, cohort A). These patients were classified as treatment failures. One additional patient died from colitis and sepsis following receipt of combination therapy and was classified as a treatment failure (Figure 1). Two other patients treated on cohort A, including one who achieved an MRD-negative CR, stopped therapy early due to combination therapy toxicity. Patients received one (n=29), two (n=10), three (n=1), or four (n=1) 28-day cycles of treatment. Details are located in Supplementary Tables 1 and 2. Four patients had CNS involvement and received protocol-specified intensified intrathecal therapy.

Four patients had primary refractory AML, one had relapsed mixed-phenotypic acute leukaemia, 27 had relapsed AML without prior salvage therapy, and 13 had relapsed disease refractory to prior salvage chemotherapy (Table 1, Supplementary Table 1). Patients with relapsed AML refractory to salvage had received one, two, or four previous lines of therapy prior to treatment. Sixteen patients had undergone HSCT prior to enrolment, including three with two previous transplantations. Thirty-one patients were treated for first relapse (including two who relapsed before completing primary therapy), eight for second relapse, and two for third relapse.

Dose reduction or discontinuation was not required due to therapy-related toxicity in cohort C. Patients receiving venetoclax with high-dose cytarabine combined with azacitidine 75mg/m² for 7 days experienced no dose limiting toxicities during rolling-6 evaluation.

Cycle 1 response was evaluable in 44 patients. Complete response with (N=24) or without (N=1) haematological recovery was achieved in 25 patients (56.8%; 95% confidence interval 41.1-71.7%), including 19 MRD negative remissions. When analysed by treatment cohort, CR/CRi was achieved in 12/22 (54.5%; 32.2-75.6%) of patients treated in cohort A, 8/15 (53.3%; 26.5-78.7%) in cohort B, and 5/7 (71.4%; 29.4-96.3%) in cohort C (Table 3). Partial responses were observed in four patients (9.1%; 2.5-21.7%), resulting in overall response rate of 29/44 (65.9%; 50-79.6%), 15/22 (68.2%; 45.1-86.1%), 9/15 (60%; 32.9-83.6%), and 5/7 (71.4%; 29.1-96.3%) in the overall study and three cohorts, respectively. Fifteen patients (34.1%; 20.5-49.9%) had no response or were treatment failures. Post-hoc analyses evaluated response in specific patient subsets. Six of 13 (46.2%; 19.2-74.9%) evaluable patients who were refractory to previous salvage therapy achieved complete responses with one additional patient achieving a partial response. One of four primary refractory patients achieved a CR and one achieved a PR. Complete response with or without hematologic recovery were observed in patients refractory to one (n=4), two (n=1), and four (n=2) previous regimens. A partial response was seen in one patient refractory to two previous regimens (Table 3, Supplementary Table 1). Allogeneic HSCT occurred in 22/25 (88%) patients who achieved a CR/CRi. All patients who were MRD-negative after cycle 1 and proceeded directly to transplant remained MRD-negative at the time of transplant.

Table 3:

Cycle 1 Response by Dose Level

	Evaluable Patients	CR/CRi*	MRD negative	PR	ORR	NR
Cohort A	22	12 (54.5%)	11 (50%)	3 (13.6%)	15 (68.2%)	7 (31.8%)
Cohort B	15	8 (53.3%)	5 (33.3%)	1 (6.7%)	9 (60%)	6 (40.0%)
Cohort C	7	5 (71.4%)	3 (42.9%)	0 (0.0%)	5 (71.4%)	2 (28.6)

Abbreviations: CR, complete response; CRi, complete response with incomplete haematological recovery (N=1 in cohort A only); MRD, minimal residual disease; PR, partial response; ORR, overall response rate (CR+CRi+PR); NR, no response; h, hour

Death following combination therapy or removal from study prior to completion of combination therapy was classified as treatment failure.

MRD negative (<0.1%) rates are shown as a percentage of all treated patients.

Recurrent genetic lesions in this relapsed/refractory population are summarized in Table 1. Responses were observed across multiple genetic subgroups, including patients with favourable-risk alterations, KMT2A rearrangements, and a variety of adverse-risk lesions. Among patients with favourable genetic alterations, the majority achieved a response, predominantly CR/CRi. Responses were also observed in patients with KMT2A rearrangements and FLT3 alterations, although outcomes were heterogenous across subgroups. Detailed distributions of responses by genetic subgroup and treatment cohort are provided in Supplementary Table 3. Because genomic alterations were not mutually exclusive, some patients are represented in more than one lesion category.

BH3 dependency profiling was performed for 24 patients with evaluable window and cycle 1 responses. BCL-2 dependence was observed in 17 patients, while seven patients demonstrated dependence on other BH3 family proteins, including BFL1 (BCL2A1) and MCL1. Of those with alternative primary dependencies, two patients had secondary BCL-2 dependence (Supplementary Table 5). Among BCL-2-dependent patients, 14 of 17 (82.4%) had a good window response and 13 (76.5%) achieved a complete response with or without hematologic recovery. Of the seven patients with alternative BH3 family dependencies, five (71.4%) had a good window response, three (42.9%) achieved a complete response, one (14.3%) had a partial response, and three (42.9%) had no response. Among the two patients with secondary BCL-2 dependence, one achieved a complete response and one a partial response; both had good window response to venetoclax. Response rates were not significantly higher among BCL-2 dependent patients; however, these findings are limited by small sample size.

Quality of life as measured by PROMIS largely remained within normative ranges without significant change from the start of therapy to the end of cycle one. Hopefulness was maintained throughout therapy in most patients (Supplementary Table 6).

Patients experienced a total of 76 non-hematologic grade 3 or higher toxicities during their first cycle of therapy; there were 23 (52.3%) patients with febrile neutropenia, 11 (25%) identifiable infections or sepsis, and 14 (31.8%) gastrointestinal disorders (Table 2). Additional infectious complications include a hepatic infection, mucosal infection, and sinusitis. Among gastrointestinal disorders, grade 3 or 4 colitis occurred in 4 patients. Other adverse events included diarrhoea, nausea/vomiting, AST/ALT elevations, and rectal haemorrhage. There were two episodes of tumour lysis, with one patient experiencing tumour lysis syndrome attributed to both venetoclax and rapid disease progression, leading to removal from the study prior to combination therapy (Table 2). One death due to sepsis and colitis, described above, was the only death possibly related to treatment. Prolonged QT was not observed although routine screening after therapy initiation was not performed. Among patients receiving more than one cycle of therapy, toxicity in subsequent cycles was consistent with the overall cohort during cycle 1 (Supplementary Table 2).

Table 2:

Patients with Non-Haematological Toxicities Possibly Related to Venetoclax During Cycle 1 by Dose Level

	Cohort A (n=22)				Cohort B (n=15)			Cohort C (n=7)
	Grade 1/2	Grade 3	Grade 4	Grade 5	Grade 1/2	Grade 3	Grade 4	Grade 1/2	Grade 3	Grade 4
Blood and Lymphatic System Disorders
DIC	0	1 (4.5%)	0	0	0	0	0	0	0	0
Febrile Neutropenia	0	13 (59.1%)	0	0	0	7 (46.7%)	0	0	3 (42.9%)	0
Cardiac Disorders
Cardiac Arrest	0	0	1 (4.5%)	0	0	0	0	0	0	0
Sinus Tachycardia	0	1 (4.5%)	0	0	0	0	0	0	0	0
Hypertension	0	0	0	0	1 (6.7%)	1 (6.7%)	0	0	0	0
Hypotension	0	0	0	0	0	2 (13.3%)	0	0	0	0
Gastrointestinal Disorders
Colitis	0	2 (9.1%)	0	0	0	2 (13.3%)	0	0	0	0
Dental Caries	0	0	0	0	0	1 (6.7%)	0	0	0	0
Diarrhoea	0	0	0	0	2 (13.3%)	0	0	1 (14.3%)	1 (14.3%)	0
Nausea	2 (9.1%)	0	0	0	1 (6.7%)	1 (6.7%)	0	0	0	0
Vomiting	2 (9.1%)	0	0	0	2 (13.3%)	1 (6.7%)	0	0	0	0
Rectal Haemorrhage	0	1 (4.5%)	0	0	0	0	0	0	0	0
Typhlitis	0	1 (4.5%)	0	0	0	0	0	0	0	0
ALT Elevation	0	0	0	0	0	1 (6.7%)	0	0	0	0
AST Elevation	0	0	0	0	0	1 (6.7%)	0	0	0	0
Abdominal Pain	0	0	0	0	0	0	0	0	1 (14.3%)	0
Other	0	1 (4.5%)	0	0	1 (6.7%)	0	0	0	0	0
Infections
Catheter-related Infection	0	1 (4.5%)	0	0	0	0	0	0	0	0
Hepatic Infection	0	1 (4.5%)	0	0	0	0	0	0	0	0
Mucosal Infection	3 (13.6%)	0	0	0	0	1 (6.7%)	0	0	0	0
Sepsis	0	0	0	1 (4.5%)	0	0	2 (13.3%)	0	0	0
Sinusitis	0	0	0	0	0	1 (6.7%)	0	0	0	0
Skin Infection	0	0	0	0	0	1 (6.7%)	0	0	0	0
Lip Infection	0	0	0	0	1 (6.7%)	0	0	0	0	0
Lung Infection	0	0	0	0	1 (6.7%)	0	0	0	0	0
Other	0	1 (4.5%)	0	0	1 (6.7%)	1 (6.7%)	0	0	2 (28.6%)	0
Metabolism and nutrition disorders
Anorexia	1 (4.5%)	0	0	0	0	1 (6.7%)	0	0	0	0
Hyperglycaemia	0	1 (4.5%)	0	0	0	0	0	0	0	0
Hyperkalaemia	0	1 (4.5%)	0	0	0	0	0	0	0	0
Hypokalaemia	0	1 (4.5%)	0	0	0	0	0	0	0	0
Tumour Lysis Syndrome	0	1 (4.5%)	0	0	0	1 (6.7%)	0	0	0	0
Dehydration	1 (4.5%)	0	0	0	0	0	0	0	0	0
Hypoalbuminemia	0	0	0	0	1 (6.7%)	0	0	0	0	0
Nervous System Disorders
Syncope	0	2 (9.1%)	0	0	0	0	0	0	0	0
Renal and Urinary Disorders
Acute Kidney Injury	0	1 (4.5%)	0	0	0	0	0	0	0	0
Urinary Retention	1 (4.5%)	0	0	0	0	0	0	0	0	0
Respiratory, Thoracic, and Mediastinal Disorders
Pulmonary Oedema	0	0	0	0	0	1 (6.7%)	0	0	0	0
Pulmonary HTN	0	0	0	0	0	1 (6.7%)	0	0	0	0
Respiratory Failure	0	0	1 (4.5%)	0	0	0	0	0	0	0
Other
Periorbital Oedema	0	1 (4.5%)	0	0	0	0	0	0	0	0
Multi-organ Failure	0	0	0	1 (4.5%)	0	0	0	0	0	0
Conjunctivitis	1 (4.5%)	0	0	0	0	0	0	0	0	0
Fever	1 (4.5%)	0	0	0	0	0	0	0	0	0
Pain	0	0	0	0	0	1 (6.7%)	0	0	0	0
Pustular Rash	1 (4.5%)	0	0	0	0	0	0	0	0	0

Abbreviations: DIC, disseminated intravascular coagulation; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HTN, hypertension

Post-hoc analysis showed that, with median follow-up of 5.3 years (IQR 4.2-6.3 years), 3-year overall survival in the entire cohort was 39.8% (95% confidence interval: 25.3-53.9%). Three-year survival in cohort A, cohort B, and cohort C was 48% (26-67%), 13.3% (2.2-34.6%), and 71.4% (25.8-92%), respectively (Supplementary Figure 3). Death was due to primary disease in 7, 12, and 2 patients in the respective cohorts.

In post-hoc analyses, we evaluated response to the window of single-agent venetoclax in thirty-one cohort A and B patients who were evaluable with available MRD at both timepoints and 30 also had evaluable end-of-therapy responses. Due to combination therapy with azacitidine, patients in cohort C were not evaluable for this analysis. Two additional patients were not evaluable as they required initiation of HDAC prior to day 8. Among 23 (74.2%; 55-88%) patients with good window response, 18 (78.2%; 56.2-92.5%) had a complete response with or without complete hematologic recovery, three (13%: 2.8-33.6%) had a partial response, and two (8.7%; 1.1-28%) had no response. Among eight (25.8%; 11.9-44.6%) patients with a poor window response, only two (25%; 3.2-65%) achieved a complete response with 62.5% (24.4-91.5%) having no response at the end of therapy (Supplementary Table 4).

Discussion

In this phase 1 expansion cohort, venetoclax combined with high-dose cytarabine at the RP2D achieved a CR/CRi in 56.8% of children, adolescents, and young adults with relapsed/ refractory AML, most of whom were MRD-negative. This combination achieved a CR/CRi in more than half of treated patients either as a two-drug combination or when combined with either idarubicin or azacitidine with acceptable safety profiles in each cohort. This report significantly expands assessments of efficacy across the genomically diverse relapsed AML landscape compared to prior reports.^13,14 In this large cohort treated at the RP2D, the venetoclax monotherapy window was active, with 74.2% of patients achieving at least a 50% reduction in circulating peripheral blasts and 71.4% of these good responders achieved a CR/CRi. This suggests that pre-therapy venetoclax resistance may predict treatment failure of venetoclax-based combination therapy in relapsed AML. Although this venetoclax monotherapy window differs from other commonly used venetoclax-based regimens in AML, these findings suggest that early venetoclax sensitivity may provide clinically relevant information. However, the routine incorporation of a venetoclax monotherapy run-in in future regimens cannot be definitively recommended based on the current data and is being further evaluated the ongoing randomized paediatric phase 3 relapsed AML study ITCC-101/APAL2020D (NCT05183035). Toxicities were manageable and consistent with those expected from intensive therapy in this population. These findings indicate that venetoclax, when combined with intensive chemotherapy, is an active treatment option for paediatric patients with relapsed/refractory AML with tolerable toxicity, supporting further investigation in this high-risk patient population.¹⁵

Other recent paediatric trials, such as the Children’s Oncology Group (COG) study AAML07P1 and the Therapeutic Advances in Childhood Leukaemia and Lymphoma Consortium’s studies have reported ORRs of 48-75% for salvage regimens in relapsed/refractory AML.^19,20 Notably, the 65.9% ORR at the RP2D in our study exceeds those rates in a similar relapsed/refractory population. Despite including more heavily pretreated patients, our results are also relatively comparable to the COG AAML1421 study evaluating CPX-351 in patients in first relapse, which had a CR/CRi rate of 76% after cycle 1, although definitions of CR vary between trials, limiting direct comparisons.^21,22 Venetoclax has been extensively studied in the adult AML population, particularly among those unfit for intensive chemotherapy. A recently published meta-analysis demonstrated improved CR/CRi rates and longer OS in 1,232 adults with previously untreated AML who received venetoclax in combination with hypomethylating agents or low-dose cytarabine, compared to chemotherapy alone.²³ A similar analysis in 222 relapsed/refractory patients showed modestly improved outcomes in adults with relapsed/refractory AML receiving venetoclax-containing regimens.²⁴ Given these modest improvements in response rates, ongoing trials are evaluating novel agents in combination with venetoclax or higher doses of venetoclax in the relapsed/refractory adult patient population.^25,26 Several trials are also exploring new combinations containing venetoclax in upfront therapy.^25,27-30 Combinations of venetoclax with a variety of conventional chemotherapy backbones in adult patients with newly diagnosed AML have produced higher rates of CR and MRD-negative remissions than intensive conventional chemotherapy alone.^27-30 Experience with venetoclax in the paediatric and young adult population is more limited. Badawi and colleagues developed a pharmacokinetic model using data from over 120 paediatric patients with relapsed/refractory malignancies across three phase I trials, including patients from the current trial, identifying an adult-equivalent of 600 mg of venetoclax as optimal for combination with intensive chemotherapy in relapsed AML patients.¹⁷ Several Phase I studies evaluating venetoclax in paediatric patients with AML have been published, including the dose-escalation phase of our study.^13,14 One multicentre study included venetoclax monotherapy and venetoclax combinations with either hypomethylating agents or cytarabine, reporting an ORR of 24% across all patients, with ORRs of 44% for high-dose cytarabine combinations and 21% for azacitidine.¹⁴ In our current cohort, the ORR in patients treated with venetoclax, cytarabine, and azacitidine suggests this combination may offer an efficacious and tolerable option for relapsed AML, though definitive conclusions are limited by the small sample size and differences in baseline patient characteristics among cohorts, including prior anthracycline exposure and clinical/ genomic risk profile such as all three NUP98 rearranged patients being in cohort B. While the durability of these responses is not well assessed in this trial due to most responding patients proceeding to stem cell transplant, our results suggest this therapy is able to achieve disease control while avoiding toxicities which preclude subsequent transplant. Other limitations of our study include a lack of randomization and modest populations of rare genetic subgroups to precisely assess responses in these groups.

Although patients with favourable genetic features exhibited the highest rates of response, encouraging results were also seen in those with adverse genetic features. Complete responses were achieved in patients with various adverse genetic risk groups including FLT3 mutations. During the dose-escalation phase, none of the five patients with activating FLT3 mutations or internal tandem duplications responded to therapy. However, definitive conclusions regarding outcomes within specific genetic subgroups remain limited by small sample sizes in these subpopulations. As enrolment expanded, broader representation of molecular subtypes within high-risk AML likely contributed to the observed heterogeneity in response, further highlighting the challenges of interpreting subgroup-specific outcomes in limited sample sizes.

Conclusion

These data further support the activity and tolerability of venetoclax-based regimens incorporating a high-dose cytarabine backbone in children, adolescents, and young adults with relapsed or refractory AML. Given differences in treatment composition and patient selection across cohorts, conclusions regarding the relative benefit of additional agents, including idarubicin or azacitidine, cannot be definitively drawn from this phase 1 dose expansion study. Combination therapy response was strongly associated with response to venetoclax monotherapy. No clear association between BH3 dependency and clinical response could be established in this cohort; further evaluation of this test in a larger cohort is indicated. Overall, these results support the ongoing investigation of venetoclax in combination with intensive chemotherapy in both children with newly diagnosed (NCT05955261) and relapsed paediatric AML (NCT05183035).

Research in context.

Evidence before this study

We searched PubMed for clinical reports published in English between January 1, 1996, and December 19, 2025, using the terms “acute myeloid leukaemia” (AML), “venetoclax”, “paediatric”, and “children”. We excluded pre-clinical studies, reviews, pre-print manuscripts, and individual patient case reports. We identified 15 clinical reports of venetoclax use in paediatric AML. Three studies, including the phase 1 portion of the present trial, evaluated venetoclax combined with intensive chemotherapy in a total of 49 children with relapsed or refractory AML. The remaining reports consisted primarily of small paediatric-specific case series with heterogeneous treatment approaches or studies in newly diagnosed disease, many of which described notable responses in heavily pretreated or multiply relapsed patients.

Added value of this study

Our results further show that venetoclax combined with high-dose cytarabine-based regimens is active and tolerable in children with relapsed AML. These data also demonstrate activity across genomic subgroups commonly observed at relapse. To our knowledge, this represents the largest report of children treated with venetoclax and intensive chemotherapy for relapsed AML and provides comprehensive efficacy analysis of patients treated at the RP2D, including those enrolled in a newly introduced expansion cohort. This study further supports use of this regimen in this population.

Implications of all available evidence

Venetoclax combined with intensive chemotherapy is active with acceptable toxicity in children, adolescents, and young adults with relapsed or refractory AML. These findings support ongoing phase 3 evaluation of this approach in paediatric relapsed AML.

Data sharing

Trial data is posted on https://clinicaltrials.gov/ as required. Reasonable requests for anonymized trial data may be directed to the corresponding author.