A phase 3 study of intensive chemotherapy with or without dasatinib in core-binding factor acute myeloid leukemia

Key Points

• •The addition of dasatinib to chemotherapy did not improve event-free and overall survival in patients with CBF-AML.
• •The addition of dasatinib to chemotherapy was associated with increased toxicity.

Visual Abstract

Abstract

Core-binding factor acute myeloid leukemia (CBF-AML) is associated with KIT mutations and deregulated expression of KIT. We report results from the randomized, open-label, phase 3 trial of intensive chemotherapy with or without the multikinase inhibitor dasatinib in adult patients with CBF-AML. Patients received “3+7” induction therapy, followed by 4 cycles of high-dose cytarabine; in the investigational arm, patients received dasatinib 100 mg daily on days 8 to 21 in induction, and on days 6 to 28 in consolidation cycles, followed by 12-month single-agent dasatinib 100 mg daily. Primary end point was event-free survival (EFS). Secondary end points included overall survival, relapse-free survival, and cumulative incidence of relapse. A total of 202 patients were randomly assigned to the standard arm (n = 102) and to the dasatinib arm (n = 100). Median age was 49 years (range, 18-77); 94 patients had t(8;21), 108 had inv(16)/t(16;16); and 58 (28.7%) patients had a KIT comutation. There was no statistically significant difference in EFS (hazard ratio, 0.92; 95% confidence interval, 0.63-1.33; P = .66) or secondary end points between treatment arms. There was also no significant difference in EFS in subgroup analyses according to age, CBF-AML type, and KIT mutation status. The incidence of serious adverse events was higher in the investigational arm (64%) than in the standard arm (36%). In patients with CBF-AML, the addition of dasatinib to intensive chemotherapy failed to improve survival outcomes. The addition of dasatinib was associated with an increase in toxicity. This trial was registered at www.ClinicalTrials.gov as NCT02013648.

The putative KIT (multikinase) inhibitor dasatinib has shown promising results for the treatment of core-binding factor acute myeloid leukemia (CBF AML) in preclinical and small pilot studies. Döhner and colleagues report the results of a randomized phase 3 prospective clinical trial with chemotherapy-fit adults with CBF AML, comparing standard induction therapy to standard induction chemotherapy plus dasatinib. The addition of dasatinib to chemotherapy failed to improve event-free and overall survival in patients with CBF AML and appeared to increase toxicity. This negative study underscores the importance of randomized prospective clinical trials and puts to rest the use of dasatinib for CBF AML.

Introduction

Acute myeloid leukemia (AML) with t(8;21)(q22;q22.1)/RUNX1::RUNXT1 and AML with inv(16)(p13.1q22)/t(16;16)(p13.1;q22)/CBFB::MYH11, commonly referred to as core-binding factor (CBF) AMLs, are distinct disease entities in the current AML classifications.^1,2 On the molecular level, both leukemias are very heterogeneous with various secondary chromosomal abnormalities and secondary gene mutations.^{3, 4, 5, 6} In the 2022 European LeukemiaNet genetic risk classification, CBF-AML are categorized in the favorable-risk group independent of the presence of secondary genetic lesions.⁷

Mutations in KIT have been identified in approximately one-third of patients with CBF-AML.^{3,4,6,8, 9, 10, 11, 12} The KIT gene, located at chromosome band 4q11-12, encodes a 145-kD transmembrane glycoprotein, which is a member of the type 3 receptor tyrosine kinase family.^13,14 After ligand binding, the receptor dimerizes, becomes phosphorylated, and activates downstream signaling pathways involved in proliferation, differentiation, and survival. Ligand-independent activation of KIT can be caused by gain-of-function mutations that are associated with CBF-AML and other human malignancies. In CBF-AML, KIT mutations cluster most frequently within exon 17 that encodes the KIT activation loop (A-loop) in the kinase domain, and in exon 8, which encodes an evolutionary highly conserved region in the extracellular portion of the KIT receptor. KIT mutations, in particular exon 17 mutations in AML with t(8;21), have been associated with inferior outcome.^{4, 5, 6,8, 9, 10, 11, 12}

In addition to KIT mutations, CBF-AML are characterized by a significantly higher KIT expression when compared with other AML subgroups.^{15, 16, 17, 18} Taken together, these findings provided a rationale for the evaluation of a KIT inhibitor not only in KIT mutant but in all CBF-AML.

In vitro studies supported the concept showing that exposure to TKIs inhibits growth of cells expressing wildtype (wt) KIT or various KIT mutants.^{19, 20, 21, 22, 23} Dasatinib is an ATP-competitive, dual SRC/ABL inhibitor. In a study by Schittenhelm et al,²¹ dasatinib was shown to potently inhibit wt KIT with a half maximal inhibitory concentration (IC₅₀) of 5 to 10 nmol/L for inhibition of autophosphorylation and cellular proliferation. Dasatinib also potently inhibited KIT juxtamembrane domain mutations with an IC₅₀ of 1 to 10 nmol/L. Notably, dasatinib is a potent inhibitor of KIT A-loop mutants, with IC₅₀ values for inhibition of autophosphorylation of KIT D816 mutants in the range of 10 to 100 nmol/L. The potency seems to be differentially influenced by various A-loop mutations, that is, KIT D816Y is 10-fold more sensitive to dasatinib than KIT D816V. The rationale to target KIT in the treatment of CBF-AML is further supported by a murine study, in which leukemic cells coexpressing RUNX1::RUNX1T1 and the A-Loop KIT N822K mutant were injected into sublethally irradiated mice, and the animals were subsequently treated with cytarabine and/or the TKI dasatinib.²⁴ The combination of cytarabine with dasatinib significantly prolonged the survival of the animals when compared with the treatment with both agents as single drugs. Beyond the inhibition of KIT, dasatinib has been shown to enhance the sensitivity of leukemic blasts to cytotoxic agents.²⁵

Two trials previously investigated the addition of dasatinib to intensive induction and consolidation chemotherapy followed by a 12-month maintenance phase in CBF-AML.^26,27 Both trials reported 3- or 4-year overall survival (OS) rates of ∼75%, and the therapy was associated with a favorable safety profile with no unexpected toxicities.

Here, we report the results of the phase 3, randomized, open-label AML Study Group (AMLSG) 21-13 trial, which investigated the addition of dasatinib to intensive induction and consolidation chemotherapy followed by a 12-month maintenance phase in patients with CBF-AML.

Material and methods

Study design and patients

AMLSG 21-13 was a randomized, open-label, phase 3 study conducted at 55 hospitals in Germany and Austria (supplemental Materials, available on the Blood website). Participants aged ≥18 years with newly diagnosed CBF-AML were eligible. Diagnoses included de novo AML, secondary AML after an antecedent myeloid neoplasm, and therapy-related AML. Patients were considered fit for intensive chemotherapy (IC) and had an Eastern Cooperative Oncology Group performance status score of 0 to 2. A full list of inclusion and exclusion criteria is available in the supplemental Materials.

All patients provided written informed consent. The AMLSG 21-13 trial was approved by the ethics committees at all sites and registered at clinicaltrialsregistry.eu (Eudra-CT number 2013-003117-18) and ClinicalTrials.gov (NCT02013648).

Treatment

Short-term treatment with hydroxyurea for up to 5 days to control hyperleukocytosis was allowed before treatment on the study protocol. Induction therapy consisted of daunorubicin 60 mg/m² per day on days 1 to 3 and cytarabine 200 mg/m² per day by continuous IV infusion on days 1 to 7; patients who failed to achieve a complete remission (CR) or CR with incomplete hematologic recovery (CRi) after the first induction cycle could receive a second cycle on protocol with daunorubicin 50 mg/m² on days 1 to 3 and cytarabine 200 mg/m² per day by continuous IV infusion on days 1 to 5. In patients with CR/CRi after induction therapy, postremission therapy consisted of 4 courses of high-dose cytarabine 3000 mg/m² every 12 hours on days 1, 3, and 5 administered over 3 hours; patients aged >60 years received an attenuated dose with cytarabine 1000 mg/m² every 12 hours on days 1, 3, and 5. In the investigational arm, dasatinib 100 mg orally once daily was added on days 8 to 21 in induction cycles and on days 6 to 28 in consolidation cycles; after completion of postremission therapy, patients in CR/CRi continued on maintenance therapy with dasatinib 100 mg orally once daily for 12 months or until relapse. Patients who were unable to complete all 4 high-dose cytarabine cycles because of toxicity were still eligible for maintenance therapy with dasatinib. Dose modifications of dasatinib due to hematologic and/or nonhematologic toxicity are given in the supplemental Materials. Due to a shortage of daunorubicin announced by the main supplier, in March 2019 the protocol was amended to allow for the administration of idarubicin (12 mg/m² per day on days 1, 3, and 5 in the first induction cycle and 10 mg/m² on days 1 and 3 in a second induction cycle) in case daunorubicin was not available.

Allogeneic hematopoietic cell transplantation from a matched related or matched unrelated donor was only recommended for patients with molecular disease persistence during consolidation therapy and during early follow-up and patients who experienced a molecular relapse.

Molecular analysis

Rapid molecular screening (within 48-72 hours) was performed centrally within our AMLSG BiO Registry in 1 of 2 AMLSG reference laboratories.²⁸ The diagnosis of CBF-AML was established by polymerase chain reaction (PCR)-based detection of RUNX1::RUNX1T1 fusion transcript resulting from t(8;21)(q22;q22.1) (or a variant form) or of CBFB::MYH11 fusion transcript resulting from inv(16)(p13.1q22) or t(16;16)(p13.1;q22). Once a diagnosis of CBF-AML was confirmed, the patient could then sign the trial-specific informed consent form. Patients could not start chemotherapy before the results were available.

Samples were also analyzed for KIT mutations (exons 8 and 17), FLT3 internal tandem duplications, and FLT3 tyrosine kinase domain mutation (codons D835/I836), as previously described.^29,30 Assessment of MRD for RUNX1::RUNX1T1 and CBFB::MYH11 fusion transcripts by real-time quantitative PCR (RQ-PCR) was performed as previously described.^31,32

Outcomes

The primary end point of the trial was event-free survival (EFS). Of note, definition of treatment failure included both hematologic as well as molecular response criteria as assessed by RQ-PCR for the CBF-AML associated fusion transcripts. For EFS analysis, treatment failure was defined as failure to achieve at least a partial remission after the first and a hematologic CR or CRi after an optional second induction cycle, hematologic relapse, molecular persistence, molecular relapse, death by any cause, whichever came first. Day 1 after randomization was assigned as the event date for patients with failure to achieve a hematologic remission by the end of induction therapy; molecular relapse and molecular persistence were assigned as events at the date of event. In an exploratory analysis, EFS was evaluated using an alternative definition in which the definition of EFS did not include molecular persistence or molecular relapse as events.

Molecular persistence was defined by RQ-PCR positivity in all bone marrow samples analyzed during consolidation therapy and by an insufficient reduction of transcript levels of <1-log during consolidation therapy. Molecular relapse was defined by (1) conversion from RQ-PCR negativity to RQ-PCR positivity in the bone marrow and/or peripheral blood sample in at least 2 consecutive bone marrow and/or peripheral blood samples obtained within 4 weeks; or (2) increase of the transcript level in bone marrow and/or peripheral blood samples of >1-log in at least 2 consecutive bone marrow and/or peripheral blood samples obtained within 4 weeks.

Secondary end points included OS, relapse-free survival (RFS), and cumulative incidence of relapse (CIR) and death (CID). RFS was defined as the time from first CR/CRi until hematologic/molecular relapse or molecular persistence or death, whichever came first. Safety end points were rate of early death/hypoplastic death; as well as type, frequency and severity of adverse events (AEs; graded using the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.03); and timing and relatedness of nonhematologic toxicity observed during different treatment cycles.

Randomization

Patients were randomized in a 1:1 ratio between the standard arm and the investigational arm. Randomization was performed at the AMLSG Clinical Trials Office with stratification according to age (18-60 years vs >60 years) and type of CBF-AML [t(8;21) vs inv(16)] in a ratio of 1:1 with equivalent allocation to the 2 arms with randomization using blocks of a defined length.

Statistical analysis

The final required number of patients was established for EFS (assumptions for sample size calculation, see the supplemental Materials). The analyses of the primary and secondary end points are based on the intention-to-treat (ITT) population, which included all randomized patients with a signed informed consent; not included in the ITT population are patients who withdrew informed consent before start of treatment.

The primary end point EFS was analyzed using a log-rank test stratified for age group and type of CBF-AML. The hazard ratio (HR) for EFS, along with the 95% confidence interval (CI), was estimated using a stratified Cox model with treatment arm as a risk factor and strata factors at randomization as variables. To evaluate the robustness of the primary end point analysis, additional stratified log-rank tests and unstratified Cox regression models as well as subgroup analyses for KIT mutation status, CBF type, age group, and sex were performed. Multivariable Cox proportional hazards models were fitted using age (10-year increase), sex, Eastern Cooperative Oncology Group performance status (0-1 vs ≥2), type of CBF-AML type, history of AML (de novo vs secondary/therapy-related), white blood cell count (WBC; log-10 transformed), KIT mutation status, FLT3 mutation status, as well as MRD status after 2 treatment cycles as time-dependent covariable. The selection of confounders included in the multivariable models was based on clinical expert knowledge. In addition, EFS was analyzed using the alternative definition. Missing values are imputed using the algorithm of substantive model compatible fully conditional specification using the R package “smcfcs” using 20 imputed data sets.^33,34 The results of the imputed data sets are pooled using the Rubin rules.³⁵

Secondary end points were analyzed in an exploratory manner. Multivariable Cox proportional hazards for OS, RFS, CIR, and CID were used by adjusting for the above-mentioned variables.

According to the study protocol, a futility interim analysis was performed in December 2018 based on 139 patients. The study should be stopped if the conditional power at interim was <30%. The conditional power under design effect given n = 139 of the patients were recruited was 37.6%, hence it was decided to continue the study.

For testing differences in baseline characteristics between treatment arms, the Fisher exact test was used for categorical data, and the Mann-Whitney test for 2-group comparisons of continuous data. Statistical analyses were performed with the statistical software environment R, version 4.4.1, using the R packages survival, version 3.7-0, smcfcs version 1.9.2, and riskRegression, version 2023.12.21.

Results

Patient and disease characteristics

Between August 2014 and February 2021, 204 patients were accrued. Patient disposition is shown in Figure 1. The ITT population comprised 202 patients (2 patients without valid informed consent were excluded), 102 patients were randomized to the standard arm and 100 patients to the dasatinib arm. The median age was 48.9 years, 108 patients had inv(16)/t(16;16) and 94 patients t(8;21); KIT mutations were detected in 58 of 202 (28.7%) patients (exon 8, n = 22 [10.9%]; exon 17, n = 40 [19.8%]; exon 8/17, n = 4 [2.0%]). Patient and disease characteristics are given in Table 1.

Figure 1.

CONSORT diagram. A total of 204 patients were enrolled in the study. Four patients were excluded due to violation of inclusion/exclusion criteria. Of these 4 patients, 2 had no valid informed consent and were therefore excluded from the ITT population. Thus, the ITT population comprised 202 patients, 102 patients were randomized to the standard arm and 100 patients to the dasatinib arm. In addition, 2 patients (1 on each treatment arm) were screening failures, 1 patient did not have CBF-AML, and 1 patient had an ejection fraction of 49% at baseline. Overall, 101 patients started treatment within the standard arm, and 99 patients within the investigational arm.

Table 1.

Patient and disease characteristics

	Standard arm	Dasatinib arm	Overall	P value
	n = 102	n = 100	N = 202
Age, y, median (range)	48.9 (18.3-75.7)	49.7 (18.2-77.0)	48.9 (18.2-77.0)	.638
Age group, y, n (%)				.730
18-60	82 (80.4)	78 (78.0)	160 (79.2)
>60	20 (19.6)	22 (22.0)	42 (20.8)
Sex, n (%)				.402
Male	51 (50.0)	44 (44.0)	107 (53.0)
Female	51 (50.0)	56 (56.0)	95 (47.0)
ECOG PS, n (%)				.362
0	50 (49.0)	61 (61.0)	111 (55.0)
1	40 (39.2)	33 (33.0)	73 (36.1)
2	7 (6.9)	5 (5.0)	12 (5.9)
Missing	5 (4.9)	1 (1.0)	6 (3.0)
WBC, ×109/L				.434
Median (range)	12.5 (0.5-177)	9.7 (0.3-205)	10.8 (0.3-205)
Missing	1	1	2
Hemoglobin, g/dL				.438
Median (range)	8.8 (3.6-14.0)	9.1 (4.2-15.4)	8.95 (3.6-15.4)
Missing	1	1	2
Platelets, ×109/L				.461
Median (range)	36 (5-247)	40 (6-230)	37 (5-247)
Missing	1	1	2
BM blasts, %				.168
Median (range)	53 (5-95)	60 (5-100)	59 (5-100)
Missing	7	6	13
PB blasts, %				.512
Median (range)	33 (0-82)	32 (0-90)	32 (0-90)
Missing	9	8	17
AML type, n (%)				.090
De novo	95 (93.1)	84 (84.0)	179 (88.6)
Secondary	1 (1.0)	3 (3.0)	4 (2.0)
Therapy-related	5 (4.9)	12 (12.0)	17 (8.4)
Missing	1	1	2
CBF-AML type, n (%)∗				.675
inv(16)/t(16;16)	53 (52.0)	55 (55.0)	108 (53.5)
del(7q)	4 (3.9)	6 (6.0)	10 (9.3)
+8	2 (2.0)	8 (8.0)	10 (9.3)
+21	4 (3.9)	6 (6.0)	10 (9.3)
+22	10 (9.8)	6 (6.0)	16 (15.0)
t(8;21)	49 (48.0)	45 (45.0)	94 (46.5)
del(7q)	1 (1.0)	4 (4.0)	5 (5.4)
+8	3 (2.9)	5 (5.0)	8 (8.6)
del(9q)	6 (5.9)	2 (2.0)	8 (8.6)
-Y	16 (15.7)	13 (13.0)	29 (31.2)
-X	11 (10.8)	6 (6.0)	17 (18.3)
KIT, n (%)				.439
Mutated	32 (31.4)	26 (26.0)	58 (28.7)
Exon 8	12 (11.8)	10 (10.0)	22 (10.9)
Exon 17	24 (23.5)	16 (16.0)	40 (19.8)
Exon 8 and 17	4 (3.9)	—	4 (2.0)
wt	70 (68.6)	74 (74.0)	144 (71.3)
FLT3-ITD, n (%)				.369
Yes	4 (3.9)	1 (1.0)	5 (2.5)
No	98 (96.1)	99 (99.0)	197 (97.5)
FLT3-TKD, n (%)				.821
Yes	10 (9.8)	11 (11.0)	21 (10.4)
No	92 (90.2)	89 (89.0)	181 (89.6)

BM, bone marrow; ECOG PS, Eastern Cooperative Oncology Group performance status; ITD, internal tandem duplication; PB, peripheral blood; TKD, tyrosine kinase domain.

^∗Including secondary chromosome abnormalities occurring in >5% of cases.

Efficacy outcomes: primary end point

Median follow-up, calculated by the reverse Kaplan-Meier estimator, was 63.4 months (95% CI, 56.8-68.1). In total, 114 events were observed, 60 events in the standard arm and 54 events in the investigational arm; failure of achieving CR/CRi after induction was observed in n = 4 vs n = 9, hematologic relapses were observed in n = 13 vs n = 13, molecular persistence in n = 19 vs n = 9, molecular relapses in n = 18 vs n = 18, and death from any cause in n = 6 vs n = 5 patients in the standard arm and dasatinib arm, respectively. Hematologic and molecular response rates are given in Table 2.

Table 2.

Response to induction therapy, rates of allogeneic HCT, and efficacy outcomes

	Standard arm	Dasatinib arm	All
	n = 102	n = 100	N = 202
Hematologic response to induction, n (%)
CR/CRi	98 (96.1)	91 (91.0)	189 (93.6)
CR	66 (64.7)	55 (55.0)	121 (59.9)
CRi	32 (31.4)	36 (36.0)	68 (33.7)
RD	1 (1.0)	1 (1.0)	2 (1.0)
ED	2 (2.0)	3 (3.0)	5 (2.5)
Missing∗	1 (1.0)	3 (3.0)	4 (2.0)
Molecular response†
After 2 cycles of chemotherapy
CR/CRiMRD−	36/77 (46.8%)	28/74 (37.8%)	64/151 (42.3%)
inv(16)	17/37 (45.9%)	15/39 (38.5%)	32/76 (42.1%)
t(8;21)	19/40 (47.5%)	13/35 (37.1%)	32/75 (42.7%)
At end of consolidation
CR/CRiMRD−	38/60 (63.3%)	39/50 (78.0%)	77/110 (70.0%)
inv(16)	19/30 (63.3%)	23/30 (76.7%)	42/60 (70.0%)
t(8;21)	19/30 (63.3%)	16/20 (80.0%)	35/50 (70.0%)
Allogeneic HCT, n (%)
HCT in CR/CRi‡	3 (2.9)	3 (3.0)	6 (3.0)
Matched-related donor	2 (2.0)	2 (2.0)	4 (2.0)
Matched-unrelated donor	1 (1.0)	1 (1.0)	2 (1.0)
Any HCT during disease course	28 (27.5)	29 (29.0)	57 (28.0)
Outcomes
Median EFS, mo	13.3 (10.5-75.1)	16.6 (11.7-NA)	14.1 (11.2-51.7)
1-y EFS rate, %	51 (42-62)	57 (48-68)	54 (48-62)
2-y EFS rate, %	43 (34-54)	49 (40-60)	46 (39-53)
3-y EFS rate, %	41 (32-52)	46 (36-57)	43 (37-51)
4-y EFS rate, %	41 (32-52)	44 (35-56)	42 (36-50)
Median OS, mo	NA	NA	NA
1-y OS rate, %	91 (85-97)	92 (86-97)	91 (87-95)
2-y OS rate, %	80 (72-88)	85 (78-93)	82 (77-88)
3-y OS rate, %	76 (68-85)	78 (70-87)	77 (71-83)
4-y OS rate, %	76 (68-85)	78 (70-87)	77 (71-83)
Median RFS, mo	11.6 (9.5-NA)	33.8 (12.6-NA)	17.4 (11.4-NA)
1-y RFS rate	50 (41-61)	62 (52-73)	56 (49-63)
2-y RFS rate	43 (35-55)	54 (44-65)	48 (42-56)
3-y RFS rate	42 (33-54)	50 (40-62)	46 (39-54)
4-y RFS rate	42 (33-54)	49 (39-61)	45 (39-53)

ED, early death; HCT, hematopoietic cell transplantation; MRD, measurable residual disease; NA, not available; RD, refractory disease.

^∗Two patients did not receive any therapy, 2 patients withdraw informed consent during the first treatment cycle.

^†Definition according to 2021 European LeukemiaNet recommendations for MRD analyses in AML.³⁶

^‡Six (3%) patients received a transplant in first CR/CRi: n = 4 without achieving MRD negativity; n = 1 had KIT mutation; n = 1 unknown; patients receiving an HCT after molecular or hematologic relapse are listed in “any HCT during disease course.”

There was no statistically significant difference in EFS between treatment arms (HR, 0.92; 95% CI, 0.63-1.33; stratified log-rank test P = .66) (Figure 2). Median EFS times were 13.3 (95% CI, 10.5, 75.1) months and 16.6 (11.7, not available) months for the standard and dasatinib arm, respectively; 4-year EFS rates were 41% in the standard arm and 44% in the investigational arm (Table 2). Within subgroup analyses also no significant differences in EFS were detected, according to KIT mutation status (mutated KIT, P = .77; wt KIT, P = .99), type of CBF-AML (inv(16), P = .94; t(8;21), P = .46), age group (18-60 years, P = .61; >60 years, P = .98), and sex (male, P = .83; female, P = .48; Figure 3).

Figure 2.

Primary and secondary end points. (A) EFS in the ITT population. In this EFS analysis, molecular persistence or molecular relapse are considered as treatment failure. There was no statistically significant difference in EFS between treatment arms (HR, 0.92; 95% CI, 0.63-1.33; stratified log-rank test, P = .66). (B) OS. There was no statistically significant difference in OS between treatment arms (HR, 0.93; 95% CI, 0.53-1.63; stratified log-rank test, P = .79). (C) RFS. There was no statistically significant difference in RFS between treatment arms (HR, 0.82; 95% CI, 0.55-1.21; stratified log-rank test, P = .31).

Figure 3.

Prespecified subgroup analyses for the primary end point EFS by treatment arm and according to patient subgroups. (A) According to KIT mutation status, mutated vs wt. (B) According to type of CBF-AML, inv(16) vs t(8;21). (C) According to age group, 18 to 60 years vs >60 years. (D) According to sex, male vs female.

Multivariable analyses for EFS supported this result (Table 3). Negative prognostic effects were detected for higher WBC count (HR, 2.02), presence of KIT mutation (HR, 1.94), and in trend for t(8;21)-positive AML (HR, 1.45) and MRD persistence (HR, 1.58).

Table 3.

Results of multivariable analysis for EFS, OS, and RFS

Variable∗	EFS		OS		RFS
	HR	95% CI	HR	95% CI	HR	95% CI
Dasatinib	0.99	0.67, 1.47	0.95	0.51, 1.78	0.86	0.57, 1.31
Age (10-year increase)	1.04	0.91, 1.19	1.19	0.96, 1.49	0.99	0.86, 1.14
Male sex	1.06	0.72, 1.57	0.91	0.49, 1.70	0.93	0.61, 1.42
ECOG (0-1 vs ≥2)	1.24	0.62, 2.46	1.68	0.64, 4.42	1.17	0.55, 2.51
s-/t-AML	1.73	0.89, 3.35	1.91	0.76, 4.83	1.72	0.81, 3.63
WBC (log10)	2.02	1.36, 3.01	1.37	0.79, 2.39	2.21	1.44, 3.40
t(8;21) [vs inv(16)]	1.45	0.95, 2.21	1.55	0.81, 2.96	1.57	1.00, 2.47
KIT mutation	1.94	1.28, 2.95	1.75	0.91, 3.37	2.15	1.38, 3.34
FLT3-ITD	0.75	0.22, 2.51	2.43	0.51, 11.51	0.76	0.22, 2.59
FLT3-TKD mutation	1.66	0.90, 3.04	1.28	0.45, 3.62	1.58	0.82, 3.05
MRD persistence	1.58	0.93, 2.70	1.61	0.65, 4.01	1.58	0.93, 2.70

ECOG PS, Eastern Cooperative Oncology Group performance status; ITD, internal tandem duplication; MRD, measurable residual disease; s-/t-AML, secondary/therapy-related AML; TDK, TKD, tyrosine kinase domain.

^∗The selection of confounders included in the multivariable models was based on clinical expert knowledge.

Similar results were obtained when applying the alternative definition of EFS, in which molecular disease persistence and molecular relapse were not considered as treatment failures (HR, 1.06; 95% CI, 0.70-1.62; P = .78; supplemental Figure 1).

Efficacy outcomes: secondary end points

OS

Overall, 49 (24%) patients died, 26 (25%) patients in the standard arm and 23 (23%) patients in the investigational arm. There was no statistically significant difference in OS between the 2 treatment arms (HR, 0.93; 95% CI, 0.53-1.63; stratified log rank test P = .79; Figure 2). Four-year OS rates were 76% in the standard arm and 78% in the investigational arm (Table 2). There was also no significant difference in OS in subgroup analyses according to KIT mutation status (mutated KIT, P = .39; wt KIT, P = .77), type of CBF-AML (inv(16), P = .66; t(8;21), P = .93), age (18-60 years, P = .60; >60 years, P = .75), and sex (male, P = .74; female, P = .18; supplemental Figure 2). In multivariable analysis, only presence of KIT mutation was associated in trend with inferior outcome (HR, 1.75; Table 3).

RFS

RFS was defined as the time from first CR/CRi until molecular persistence, molecular or hematologic relapse, or death, whichever came first. Overall, 101 of 189 (53%) patients had RFS events with a numerically higher rate of events in the standard arm (n = 56/98 [57%]) than in the investigational arm (n = 45/91 [49%]). The difference between the 2 treatment arms was not statistically significant (HR, 0.82; 95% CI, 0.55-1.21; stratified log-rank test, P = .31; Figure 2). The 4-year RFS rate was 42% in the standard arm and 49% in the investigational arm (Table 2). There was also no significant difference in RFS in subgroup analyses according to KIT mutation status (mutated KIT, P = .59; wt KIT, P = .64), type of CBF-AML (inv(16), P = .83; t(8;21), P = .21), age (18-60 years, P = .24; >60 years, P = .91), and sex (male, P = .40; female, 0.90; supplemental Figure 3). In multivariable analysis, negative prognostic effects on RFS were detected for higher WBC (HR for log10 increase, 2.21), KIT mutation (HR, 2.15), and presence of AML with t(8;21) (HR, 1.57; Table 3). A negative trend was observed for MRD persistence (HR, 1.58; Table 3).

CIR and CID

Analyses of CIR and CID are based on 189 patients who achieved CR/CRi by the end of induction therapy. Time to relapse was defined as the time from first CR/CRi until molecular persistence, or molecular or hematologic relapse. Overall, 90 patients experience disease relapsed and 11 patients died in CR/CRi. There was no difference in CIR and CID between treatment arm (stratified Gray test for CIR, 0.37; and CID, 0.96; supplemental Table 1; supplemental Figure 4).

Interaction tests

To assess whether the therapeutic effect of dasatinib on the survival end points (EFS, OS, RFS) differed between CBF-AML types, an interaction test was performed. For all survival end points, no statistically significant interaction effect was detected between treatment with dasatinib and type of CBF-AML (dasatinib::t(8;21): EFS, P = .345; OS, P = .391; RFS, P = .261).

Maintenance therapy

To analyze the effect of maintenance therapy, we performed a landmark analysis at the end of consolidation cycle 4 with 130 patients in hematologic remission. We found no difference between the treatment arms in RFS (44 events) and OS (21 events) using stratified log-rank tests (P value, 1.0 for RFS and 0.6 for OS; supplemental Figure 5).

Measurable residual disease

There were no statistically significant differences in the kinetics of CBFB::MYH11 and RUNX1::RUNX1T1 fusion transcript levels between treatment arms after induction and consolidation cycles, neither in inv(16) nor in t(8;21) cases (supplemental Figure 6). Similarly, molecular response rates after 2 cycles of chemotherapy and at the end of treatment were comparable between treatment arms (Table 2).

Study drug exposure and toxicity

Exposure to dasatinib

In 4 patients, dasatinib was not administered at all. In general, there was good exposure to dasatinib, with median number of days on study drug and the median cumulative dose of dasatinib being 14 days (days 8-21) and 1400 mg in induction cycle 1, 25 days (days 5-28) and 2500 mg in all consolidation cycles, and 256 days and 23.000 mg in maintenance treatment. Dose reduction, mainly due to toxicity, occurred most frequently in induction 1 (40%) and during maintenance (37%) but were mostly transient.

Thirty-one (16%) patients prematurely discontinued from the study due to adverse events, 27 patients in the investigational arm and 4 patients in the standard arm (supplemental Table 2).

AEs

AEs most frequently occurred in the MedDRA System Organ Classes blood and lymphatic system (96%), investigations (96%), gastrointestinal disorders (92%), infections and infestations (89%), and general disorders and administration site condition (87%) (supplemental Tables 4 and 5). AEs grade ≥3 occurring in ≥10% of patients or showing a difference between treatment arms are given in Table 4. In the investigational arm, colitis (8% vs 1%, P = .02), acute kidney injury (4% vs 0%, P = .04), and atrial fibrillation (4% vs 0%, P = .04) occurred more frequently. Leukopenia grade ≥3 was more prominent in the standard arm (91%) compared with the investigational arm (79%; P = .02).

Table 4.

AEs grade 3 or more occurring in at least 10% of patients or showing a difference between treatment arms

AEs	Standard arm (n = 101)	Dasatinib arm (n = 99)	P value
Hematologic, n (%)
Thrombocytopenia	95 (94)	90 (91)	.53
Leukopenia	92 (91)	78 (79)	.02
Neutropenia	65 (64)	69 (70)	.53
Anemia	89 (88)	87 (88)	1.00
Nonhematologic, n (%)
Febrile neutropenia	43 (43)	44 (44)	.85
Atrial fibrillation	0 (0)	4 (4)	.04
Colitis	1 (1)	8 (8)	.02
Diarrhea	8 (8)	14 (14)	.21
Nausea	9 (9)	15 (15)	.22
Pyrexia	13 (13)	19 (19)	.25
Device-related infection	13 (13)	8 (8)	.29
Infection	26 (26)	21 (21)	.53
Pneumonia	19 (19)	29 (29)	.09
C-reactive protein increased	8 (8)	14 (14)	.21
Hypokalemia	7 (7)	11 (11)	.31
Acute kidney injury	0 (0)	4 (4)	.04

SAEs

The proportion of patients with severe AEs (SAEs) was higher (64%) in the investigational arm than in the standard arm (36%). Forty-one percent of SAEs in the investigational arm were assessed to be related to dasatinib. The most frequent SAEs were pneumonia/pneumonitis, sepsis, febrile neutropenia, and pyrexia. This was true for both treatment arms; however, incidence of these events was higher in the investigational arm. Supplemental Figure 7 shows cumulative incidence curves according to treatment arm. For all time points, cumulative incidence of SAEs was significantly increased in the investigational arm compared with the standard arm (P < .001). There were 10 patients with fatal SAEs, 5 patients in each treatment arm. Seven patients died due to infection (sepsis, n = 3; pneumonia, n = 2; acute respiratory distress syndrome, n = 1; severe acute respiratory syndrome coronavirus 2 infection, n = 1), and 3 patients died due to multiorgan failure or death of unknown origin.

Fifty-seven patients started dasatinib maintenance treatment

During maintenance treatment, most frequent hematologic AEs were anemia (63%) and thrombocytopenia (46%), and most frequent nonhematologic AEs were diarrhea (32%) and fatigue (28%; supplemental Table 6). Of 8 patients who withdrew from the study due to AEs during maintenance, 2 had developed pleural effusion, 1 patient pulmonary edema, and 1 patient pneumonitis.

Hematologic recovery

The addition of dasatinib had no negative impact on hematologic recovery rates (supplemental Figure 8). In consolidation cycles 2 and 3, recovery of neutrophils appeared somewhat faster in the dasatinib arm.

Discussion

In this phase 3 randomized, open-label trial in patients with CBF-AML, the addition of the tyrosine kinase inhibitor dasatinib to intensive induction and consolidation chemotherapy followed by a 12-month maintenance phase failed to improve survival outcomes. Subgroup analyses according to KIT mutational status, type of CBF-AML, age, and sex also did not show any beneficial effect of dasatinib. In multivariable analysis, the presence of a KIT mutation remained a negative prognostic factor. The clinical data are supported by the results from MRD analyses showing no difference in the kinetics of fusion transcript levels after induction and consolidation cycles between treatment arms.

Two previous single-arm phase 1/2 trials investigated the combination of IC with dasatinib in CBF-AML.^26,27 In the Cancer and Leukemia Group B (CALGB) 10801 trial of 61 patients, outcomes of patients with mutated vs wt KIT (3-year OS rates 73% and 76%, respectively) were similar, suggesting that the addition of dasatinib may overcome the negative impact of KIT mutations.²⁷ In the AMLSG 11-08 trial of 89 patients, the 4-year OS rate was 74.7%, mutated KIT remained a poor prognostic factor for EFS and OS.²⁶ These outcome data appeared to be favorable when compared with historical data of a meta-analysis of randomized trials investigating gemtuzumab ozogamicin (GO; 5-year OS rate of 55% with IC alone).³⁷ However, the OS rates observed in our study are comparable with more contemporary data from patients treated with IC alone.³⁸

Two small prospective, single-arm phase 2 trials evaluated the combination of IC with midostaurin in patients with CBF-AML, 1 trial of 34 patients that included all CBF-AML,³⁹ and a second of 18 patients that was restricted to AML with t(8;21) harboring a KIT and/or a FLT3 internal tandem duplication.⁴⁰ Both trials failed to reach the prespecified primary end points of 2-year relapse incidence and 2-year EFS rate, respectively. In both trials, the relapse incidence remained high with a 2-year relapse rate of 51.5% and a 2-year EFS rate of 55.6%. A recent retrospective analysis compared real-world outcomes of 200 patients treated with IC alone, IC plus GO, and IC plus a KIT inhibitor (dasatinib, n = 14; midostaurin, n = 7).⁴¹ Notably, patients who received IC plus a KIT inhibitor had a significantly better 3-year EFS and OS compared with patients receiving IC alone or IC plus GO. However, it was not reported how patients were selected to receive a KIT inhibitor.

In the previous single-arm studies, no unexpected toxicities were reported for the addition of dasatinib to IC. However, the results from our randomized trial clearly showed an increase in toxicity in the investigational arm. AEs grade ≥3 that occurred more frequently with dasatinib were colitis, acute kidney injury, and atrial fibrillation. Gastrointestinal AEs including colitis are known side effects also observed with dasatinib monotherapy.⁴² The proportion of patients with SAEs was higher (64%) in the investigational arm than in the standard arm (36%). Most SAEs (41%) in the investigational arm were assessed to be related to dasatinib. The most frequent SAEs were pneumonia/pneumonitis, sepsis, febrile neutropenia, and pyrexia. Thirty-one (16%) patients prematurely discontinued from the study due to AEs, 27 patients in the investigational arm and 4 patients in the standard arm. Six patients discontinued dasatinib due to pleural effusion, pulmonary edema, or pneumonitis, 4 of them during maintenance therapy. Dasatinib did not negatively impact hematologic recovery rates, during consolidation cycles 2 and 3, there even appeared to be a slightly faster recovery of neutrophils, possibly owing to the differentiation-inducing effect that has been described for dasatinib in vitro and in an anecdotal case report.^{43, 44, 45}

The reason why dasatinib did not exert a beneficial effect, also in the KIT-mutated cases, remains speculative. The specific activity of dasatinib against both mutant and wt KIT may not be sufficient to translate into an outcome benefit. Preclinical studies have shown that sensitivity to dasatinib may vary between different mutants, for example, with KIT D816V being less sensitive than other A-loop mutations.²¹ However, the scarce data reported for combining IC with midostaurin are also not encouraging, questioning the further evaluation of these multikinase inhibitors in CBF-AML. Whether highly potent and more selective KIT inhibitors such as avapritinib, bezuclastinib, or elenestinib may improve outcome of KIT-mutated CBF-AML remains to be tested.⁴⁶

In conclusion, dasatinib combined with IC had no beneficial effect on outcome parameters, neither in the entire cohort nor in patient subgroups. The data do not support the encouraging data reported in single-arm phase 2 studies, highlighting again the importance of investigating new agents in controlled randomized trials. In fact, the addition of dasatinib was associated with an increase in toxicity. Based on the data from this trial, the use of dasatinib for patients with CBF-AML cannot be recommended in clinical practice.

Conflict-of-interest disclosure: H.D. reports an advisory role with AbbVie, Otsuka, Pfizer, Servier, and Syndax; reports research funding to the institution from AbbVie, Astellas, Bristol Myers Squibb (BMS), Jazz Pharmaceuticals, and Servier; and reports travel support from AbbVie and Servier. W.F. reports an advisory role with Jazz Pharmaceuticals, Celgene, MorphoSys, Ariad/Incyte, Stemline Therapeutics, Clinigen, Daiichi Sankyo, Otsuka, and Servier; reports travel/conference support from Amgen, Jazz Pharma, AbbVie, Daiichi Sankyo, and Servier; and reports support for medical writing for Amgen, Pfizer, and AbbVie. K.G. reports honoraria from BMS. E.K. reports honoraria and travel support from AbbVie, Astellas, BeiGene, Daiichi Sankyo, Jazz, and Servier. J.W. reports honoraria from Novartis, BMS, Amgen, Jazz, Sanofi, AbbVie, Astellas, Pfizer, Sobi, and Stemline Therapeutics. M.H. reports honoraria from Astellas, Daiichi Sankyo, Janssen, Miltenyi, Otsuka, Qiagen, and Servier; reports consultancy for AbbVie, AvenCell, Ascentage Pharma, BMS, Janssen, Jazz Pharmaceuticals, LabDelbert, Novartis, Pfizer, and Servier; and reports research funding to the institution from AbbVie, Bayer Pharma AG, Jazz Pharmaceuticals, Glycostem, Karyopharm, PinotBio, Servier, and Toray. M.L. reports an advisory board role with AbbVie, Astex Pharmaceuticals, Imago BioSciences, Janssen, and Otsuka; reports research support to the institution from Janssen; reports provision of clinical study drug (all-trans retinoic acid) by Cheplapharm; and reports provision for in vitro studies by Imago BioSciences and Foghorn Therapeutics. H.-J.T. reports honoraria from Sanofi-Aventis, Amgen, Delbert Pharma, AbbVie, AstraZeneca, and Oncopeptides and reports travel/conference support from Janssen, Amgen, and Sanofi-Aventis. L.L.T. reports an advisory role for Blueprint Medicines, BMS, GSK, Novartis, Pfizer, and Sobi; reports honoraria from AOP Pharma, BMS, Jazz Pharmaceuticals, Novartis, and Sobi; and reports travel support from AbbVie and AOP Pharma. L.F. reports an advisory role with AbbVie, Amgen, Kite, Jazz, Pfizer, and Janssen; reports research grants from AbbVie and Kite; and reports speakers bureau participation with Kite. A.W. reports honoraria from AbbVie, Astellas, BMS, Daiichi Sankyo, Jazz Pharmaceutical, and Novartis and reports research funding from BMS/Celgene. U.M. reports an advisory role for Astellas, BMS, AstraZeneca, Pfizer, and Roche and reports travel support from Lilly, BeOne, and Pfizer. M.R. reports honoraria from AOP Pharma, BMS, GSK, Sobi, Johnson & Johnson, Cogent, BeiGene, Lilly, and Merck. V.I.G. reports an advisory role for Jazz Pharmaceuticals, AbbVie, Boehringer-Ingelheim; reports speakers bureau participation with Pfizer, Janssen, AbbVie, and Otsuka; and reports travel support from AbbVie and Jazz Pharmaceuticals. F.T. reports an advisory role for BMS, Novartis, AbbVie, Menarini, Rigel, and Agile and reports honoraria from Jazz, Astellas, and AbbVie. P.P. reports honoraria for an advisory role and/or speakers bureau from AbbVie, Agios, Astellas, Astex, AstraZeneca, BeiGene, BMS, Celgene, Incyte, Jazz, Novartis, Otsuka, Pfizer, and Sunesis and reports travel support from AbbVie, Amgen, BeiGene, BMS, Celgene, Janssen, Novartis, and Takeda. L.B. reports honoraria from AbbVie, Amgen, Astellas, BMS, Daiichi Sankyo, Gilead, GSK, Janssen, Jazz Pharmaceutical, Miltenyi, Novartis, Otsuka, Pfizer, Roche, Sanofi, and Servier and reports research funding from Bayer and Jazz Pharmaceuticals. K.D. reports a consultancy role with honoraria with AbbVie, Daiichi Sankyo, Janssen, Jazz, Novartis, BMS, and Celgene and reports clinical research funding to the institution from Novartis, AbbVie, Astellas, BMS, Celgene, Jazz Pharmaceuticals, Kronos Bio, and Servier. The remaining authors declare no competing financial interests.

A complete list of the AML Study Group institutions and investigators appears in the supplemental Materials.