Hematopoietic Stem Cell Transplantation Outcomes for High-Risk AML: A Report From the Children's Oncology Group

Abstract

PURPOSE

Hematopoietic stem cell transplantation (HSCT) is used as consolidation for pediatric patients with high-risk AML in first complete remission (CR1). The definition of high-risk AML has evolved considerably over the past two decades with the successive identification of new unfavorable risk factors. We conducted a cross-study analysis to determine whether HSCT improves the outcomes of patients with contemporarily defined high-risk AML.

METHODS

We combined data from AAML0531 and AAML1031, the last two phase III clinical trials completed by the Children's Oncology Group (COG). These two trials established the prognostic importance of measurable residual disease (MRD) and several high-risk cryptic cytogenetic/molecular (CM) alterations, which were applied to reclassify patients in the current COG phase III clinical trial, AAML1831. We compared the outcomes after HSCT in CR1 with those after chemotherapy alone in CR1 in the redefined high-risk group.

RESULTS

Our study cohort comprised 463 patients with high-risk CM alterations and 72 patients with standard-risk (SR) CM results with positive MRD at end of induction I. In all, 33.9% and 45.8% of these groups underwent HSCT in CR1, respectively. HSCT was associated with decreased relapse and improved disease-free survival (DFS) in both groups. In the high-risk CM group, 5-year DFS was 26.0% (95% CI, 20.6 to 31.6) and 49.8% (95% CI, 41.7 to 57.4; P < .001) in patients receiving chemotherapy alone and HSCT, respectively. In the SR CM and MRD+ groups, DFS was 16.9% (95% CI, 4.3 to 36.7) compared with 50.9% (95% CI, 32.7 to 66.5; P = .032). HSCT was also associated with improvement in outcomes based on multivariable analysis and across subgroups defined by clinical trial and by high-risk CM subtype, with the exception of chromosome 7 or 5 loss.

CONCLUSION

HSCT was associated with improved outcomes in pediatric patients with contemporarily defined high-risk AML.

INTRODUCTION

The use of hematopoietic stem cell transplantation (HSCT) as consolidation therapy for AML diagnosed in children, adolescents, and young adults has been debated for decades. Before the mid-2000s, many pediatric cooperative groups used HSCT in first complete remission (CR1) for all patients with a human leukocyte antigen (HLA)-matched related donor available, regardless of disease risk. To assess the impact of disease risk on the potential benefits of HSCT, we previously performed a cross-study analysis encompassing four phase III trials that together enrolled patients from the late 1980s through the early 2000s. This study assessed the influence of disease risk on the benefit of HSCT in CR1, compared with consolidation chemotherapy alone. At that time, disease risk was defined primarily by a limited number of cytogenetic alterations identified by G-banding and fluorescence in situ hybridization. Patients with t(8;21) (or RUNX1::RUNX1T1) or inv(16) (or CBFB::MYH11) were defined as having low-risk disease, whereas patients with monosomy 7, monosomy 5, or deletion 5q were considered to have high-risk disease. Patients with other cytogenetic alterations or normal karyotypes comprised the standard-risk (SR) group, comprising well over half of the patients. Our cross-study analysis demonstrated that the benefit of HSCT was observed in patients with SR disease. A limitation of the study was the small number of patients transplanted for high-risk disease, which precluded firm conclusions being drawn regarding the effect of HSCT for this group.¹ We performed a second study, where we expanded the number of patients transplanted for high-risk cytogenetics by incorporating additional patients from the Center for International Blood and Marrow Transplant Research registry. This study also demonstrated no benefit to HSCT.²

CONTEXT

• Key Objective

• To investigate whether hematopoietic stem cell transplantation (HSCT) in first complete remission (CR1) improves outcomes for pediatric AML.

• Knowledge Generated

• For pediatric patients diagnosed with high-risk AML based on cytogenetic or molecular features, HSCT in CR1 was associated with improved outcomes compared with chemotherapy alone in CR1. Similar results were observed based on multivariable analysis and across subgroups defined by clinical trial and molecular subtype, with the exception of chromosome 7 or 5 loss.

• Relevance (C. Craddock)

• Advances in transplant technology coupled with greater availability of alternative donors continue to increase the number of children in whom transplant can be delivered with relative safety. Data from this study demonstrate improved outcomes in pediatric patients with high risk AML confirming the importance of both detailed risk stratification and prioritizing donor identification at diagnosis in this sizable patient population.*

• *Relevance section written by JCO Associate Editor Charles Craddock, MD.

Since that time, a myriad of new cytogenetic/molecular (CM) risk factors have been uncovered, largely, through the use of high-throughput sequencing. This research has identified both low-risk and high-risk factors, leading to the expansion of the low-risk and high-risk groups. A second innovation, multidimensional flow-based marrow measurable residual disease (MRD) testing performed at the end of the first cycle of induction, has been harnessed to further refine risk stratification. Using end of induction 1 (EOInd1) MRD testing, patients with SR CM alterations can be separated into the low-risk (MRD−) and high-risk (MRD+) groups.

Advances have also occurred in the approaches to unrelated adult donor, unrelated cord blood, and mismatched related donor transplants since the early 2000s, making it possible to safely expand access to HSCT in patients lacking a matched related donor.³

The Children's Oncology Group (COG) has leveraged the progress made in risk classification and HSCT to refine the use of HSCT in pediatric AML. Use of alternative donor transplants was introduced with the AAML0531 trial (ClinicalTrials.gov identifier: NCT00372593)⁴ and now comprise the majority of transplants. In the AAML1031 trial (ClinicalTrials.gov identifier: NCT01371981),⁵ COG introduced the binary risk classification system described above and limited the use of HSCT to patients in the high-risk group. With these changes, the original practice of assigning all patients with an HLA-matched related donor, regardless of disease risk, to HSCT has been replaced by one where all patients falling into the newly expanded high-risk group are assigned to HSCT with the best available donor. Importantly, with the adoption of an expanded list of new high-risk CM markers in the AAML1831 trial (ClinicalTrials.gov identifier: NCT04293562),^6-17 the high-risk group is projected to encompass approximately 40% of patients. The efficacy of HSCT for this greatly expanded high-risk group remains uncertain.

To address this matter, we combined clinical and laboratory data from the AAML0531 and AAML1031 trials. Research data (EOInd1 MRD testing for AAML0531 and high-throughput sequencing data for both studies) were incorporated to supplement protocol-mandated laboratory risk testing as needed to classify patients in the combined data set according to the AAML1831 risk schema. Outcomes of patients receiving HSCT and chemotherapy alone were compared in these high-risk patients.

METHODS

Study Design and Patients

Clinical data and biological samples were combined from patients enrolled on two consecutive COG phase III trials for de novo AML: AAML0531 (ClinicalTrials.gov identifier: NCT00372593)⁴ and AAML1031 (ClinicalTrials.gov identifier: NCT01371981).⁵ Each protocol was approved by the National Cancer Institute's central institutional review board (IRB) and the local IRB for each participating institution. Patients and/or families provided informed consent or assent as appropriate. Data for this analysis were frozen on December 30, 2019, for AAML0531 and on June 30, 2022, for AAML1031.

For AAML0531, patients (age 0-29 years) were enrolled from 2006 to 2010 and then randomly assigned to receive gemtuzumab ozogamicin in combination with conventional chemotherapy. Patients with monosomy 7, monosomy 5, deletion 5q, FLT3 with internal tandem duplication (ITD) and allelic ratio >0.4 (based on an interim amendment), or >15% blasts (based on morphologic evaluation after induction I) were assigned to the high-risk group. Patients with t(8;21) or inv(16) were assigned to the low-risk group. Those lacking high-risk or low-risk features were assigned to the SR group. Patients proceeded with HSCT in the setting of high-risk disease with an available donor (matched related, unrelated, or cord blood donor) or SR disease with an available matched related donor.

For AAML1031, patients (age 0-29 years) were enrolled from 2011 to 2017 and then randomly assigned to receive bortezomib in combination with conventional chemotherapy (comparable outcomes were observed between these two treatment arms) and nonrandomly assigned to receive sorafenib in the setting of FLT3-ITD AML (with allelic ratio >0.4). Patients with monosomy 7, monosomy 5, deletion 5q, and FLT3-ITD and allelic ratio >0.4 were assigned to the high-risk group. Patients with t(8;21), inv(16), or somatic mutations in either NPM1 or CEBPA were assigned to the low-risk group. EOInd1 MRD testing was used to assign patients with SR CM factors to the low-risk group (negative) and to the high-risk group (positive). Patients proceeded with HSCT in the setting of high-risk disease with an available donor.

An MRD threshold of ≥0.05% was used to categorize patients as MRD+ (≥0.05%) or MRD− (<0.05%).

Whole-genome sequencing and transcriptome sequencing (RNA-seq) were performed on diagnostic samples (Fig 1A; Data Supplement, Table S1, online only) as part of the National Cancer Institute (NIH) Therapeutically Applicable Research to Generate Effective Treatments Initiative, and variant/fusion calling was performed as previously described.^18,19

FIG 1.

(A) Flow diagram of COG clinical trials, CM biomarker development, reclassification based on high-throughput sequencing, and methodology. (B) Alluvial diagram comparing historical risk stratification with contemporary CM risk stratification preceding further assignment by measurable residual disease status within the study cohort. ^aRetrospective. ^bMLLT1, MLLT2, MLLT4, MLLT10, ABI1. ^c≥90 days of age at diagnosis. BM, bone marrow; CCG, Children's Cancer Group; CM, cytogenetic/molecular; COG, Children's Oncology Group; CR1, first complete remission; HSCT, hematopoietic stem cell transplantation; MRC, Medical Research Council; MRD, measurable residual disease; PB, peripheral blood; POG, Pediatric Oncology Group.

Of the 2,024 eligible patients enrolled on AAML0531 and AAML1031, 1,770 had sufficient CM and MRD data to assess risk according to the contemporary AAML1831 risk stratification schema (Data Supplement, Table S1). Of the 1,770 patients with sufficient CM and MRD data, 1,445 achieved morphologic complete remission (CR) and received the three chemotherapy courses that preceded HSCT (if prescribed) on AAML0531 or AAML1031 (Fig 2).

FIG 2.

Flow diagram of study cohort. CM, cytogenetic/molecular; CR1, first complete remission; HSCT, hematopoietic stem cell transplantation; MRD, measurable residual disease; SR, standard-risk.

Statistical Analysis

Overall survival (OS) and disease-free survival (DFS) analyses were performed using Kaplan-Meier estimates. Relapse rate was reported as cumulative incidence of relapse (CIR). OS was defined as the time from end of intensification I to death or last follow-up. DFS was defined as the time from end of intensification I to first event (relapse or death) or last follow-up. CIR was defined as the time from end of intensification I to relapse or last follow-up. Cox proportional hazards regression models were used to estimate hazard ratios (HRs) for univariable and multivariable analyses of OS, DFS, and CIR. Differences in OS and DFS between groups were determined using log-rank testing. Differences in CIR between groups were determined using Gray's test. All P values are two-sided and calculated without multiple-testing adjustments. The chi-square test was used to test the significance of differences in proportions, and Fisher's exact test was used when data were sparse. Differences in medians were compared by the Mann-Whitney test. A P < .05 was considered statistically significant.

RESULTS

Patient Characteristics

We considered the two groups comprising our contemporarily defined high-risk stratum, patients with high-risk CM alterations and patients with SR CM results and EOInd1 MRD separately. Of the 1,445 patients aggregated from the two studies, application of the AAML1831 CM risk criteria resulted in the reclassification of 24% of patients (Fig 1B). In total, 636, 346, and 463 were classified as low-, standard-, and high-risk, respectively. EOInd1 MRD results were available on 326 patients with SR CM alterations, of which 72 and 254 were MRD+ and MRD−, respectively.

In all, 33.9% of the 463 patients with high-risk CM alterations underwent HSCT in CR1 and 45.8% of the patients with SR CM results with EOInd1 MRD underwent HSCT in CR1. Only 1.6% and 4.7% of patients who were assigned CM low-risk and SR without EOInd1 MRD underwent HSCT in CR1, respectively (Fig 2). A comparison of baseline characteristics for patients receiving HSCT in CR1 and those receiving chemotherapy alone in CR1 is shown for the patients who were assigned CM high-risk and SR/MRD+ status in Table 1, respectively. HSCT information is reported in the Data Supplement (Tables S2 and S3).

TABLE 1.

Patient Characteristics for CM High-Risk and SR MRD+ Group

Characteristic	CM High-Risk			CM SR MRD+
	Chemo in CR1 (n = 306), No. (%)	HSCT in CR1 (n = 157), No. (%)	P	Chemo in CR1 (n = 39), No. (%)	HSCT in CR1 (n = 33), No. (%)	P
Gender
Female	149 (48.7)	71 (45.2)	.479	25 (64.1)	12 (36.4)	.019
Male	157 (51.3)	86 (54.8)		14 (35.9)	21 (63.6)
Race
American Indian or Alaska Native	3 (1.1)	0	.554	0	0	—
Asian	11 (4.0)	9 (6.4)	.275	2 (5.9)	1 (3.4)	1.000
Black or African American	40 (14.4)	13 (9.2)	.129	6 (17.6)	2 (6.9)	.270
Multiple races	1 (0.4)	0	1.000	0	0	—
Native Hawaiian or Pacific Islander	3 (1.1)	0	.554	0	0	—
White	219 (79.1)	119 (84.4)	.190	26 (76.5)	26 (89.7)	.170
Unknown	29	16		5	4
Ethnicity
Hispanic or Latino	52 (17.8)	21 (14.1)	.321	7 (20.0)	6 (19.4)	.948
Not Hispanic or Latino	240 (82.2)	128 (85.9)		28 (80.0)	25 (80.6)
Unknown	14	8		4	2
Age, years
0-1	83 (27.1)	20 (12.7)	<.001	7 (17.9)	4 (12.1)	.494
2-10	104 (34.0)	64 (40.8)	.151	12 (30.8)	12 (36.4)	.616
≥11	119 (38.9)	73 (46.5)	.116	20 (51.3)	17 (51.5)	.984
Study
AAML0531	126 (41.2)	67 (42.7)	.757	14 (35.9)	4 (12.1)	.020
AAML1031	180 (58.8)	90 (57.3)		25 (64.1)	29 (87.9)
Study arma
AAML0531 (arm A)	55 (18.0)	35 (22.3)	.266	7 (17.9)	1 (3.0)	.045
AAML0531 (arm B)	71 (23.2)	32 (20.4)	.490	7 (17.9)	3 (9.1)	.326
AAML1031 (arm A)	82 (26.8)	18 (11.5)	<.001	15 (38.5)	15 (45.5)	.549
AAML1031 (arm B)	81 (26.5)	26 (16.6)	.017	10 (25.6)	14 (42.4)	.132
AAML1031 (arm C)	17 (5.6)	46 (29.3)	<.001	0	0	—
MRD at end of induction I
Negative	205 (74.5)	72 (49.7)	<.001	39 (100)	33 (100)	1.000
Positive	70 (25.5)	73 (50.3)		0	0
Unknown	31	12		0	0

Abbreviations: CM, cytogenetic/molecular; CR1, first complete remission; GO, gemtuzumab ozogamicin; HSCT, hematopoietic stem cell transplantation; ITD, internal tandem duplication; MRD, measurable residual disease; SR, standard-risk.

^a AAML0531 (A = no GO; B = GO); AAML1031 (A = no bortezomib; B = bortezomib; C = sorafenib for FLT3-ITD).

Impact of HSCT on Outcomes in the High-Risk Disease Groups

In the high-risk CM group, HSCT in CR1 was associated with improved outcomes. Five-year OS was 43.7% (95% CI, 37.9 to 49.3) compared with 58.8% (95% CI, 50.5 to 66.1; P = .006) in patients receiving chemotherapy alone in CR1 (n = 306) and HSCT in CR1 (n = 157), respectively. DFS was 26.0% (95% CI, 20.6 to 31.6) compared with 49.8% (95% CI, 41.7 to 57.4; P < .001), and the CIR was 69.3% (95% CI, 63.1 to 74.7) compared with 38.0% (95% CI, 30.4 to 45.6; P < .001; Fig 3; Table 2). In multivariable analysis, after adjustment for EOInd1 MRD status and clinical trial treatment arm, the HRs for the impact of HSCT on OS, DFS, and CIR were 0.711 (95% CI, 0.52 to 0.98; P = .036), 0.518 (95% CI, 0.38 to 0.71; P < .001), and 0.386 (95% CI, 0.27 to 0.54; P < .001; Table 3), respectively. We also performed multivariable analysis, incorporating interaction terms for HSCT in CR1 and EOInd1 MRD. Within the EOInd1 MRD− group, HSCT in CR1 was associated with improved DFS and CIR compared with chemotherapy alone in CR1. Within the EOInd1 MRD+ group, HSCT was associated with improved OS, DFS, and CIR (Data Supplement, Table S4).

FIG 3.

Kaplan-Meier estimates of OS, DFS, and CIR based on contemporary CM risk stratification (standard and high) and MRD status (positive for SR), comparing patients who received chemotherapy alone versus hematopoietic stem cell transplant in first complete remission. CIR, cumulative incidence of relapse; CM, cytogenetic/molecular; DFS, disease-free survival; HSCT, hematopoietic stem cell transplantation; MRD, measurable residual disease; OS, overall survival; SR, standard-risk.

TABLE 2.

Outcomes Based on Chemotherapy Versus HSCT in CR1 Stratified by CM Risk and MRD Status

Risk and Intervention	No.	5 Year Overall Survival From End of Intensification I		5 Year Disease-Free Survival From End of Intensification I		5 Year Cumulative Incidence of Relapse From End of Intensification I
		% (95% CI)	P	% (95% CI)	P	% (95% CI)	P
CM standard-risk and MRD+
Chemo in CR1	39	39.6 (24.3 to 54.5)	.191	16.9 (4.3 to 36.7)	.032	74.3 (47.2 to 88.9)	.025
HSCT in CR1	33	53.7 (35.1 to 69.0)		50.9 (32.7 to 66.5)		37.0 (20.5 to 53.5)
CM high-risk
Chemo in CR1	306	43.7 (37.9 to 49.3)	.006	26.0 (20.6 to 31.6)	<.001	69.3 (63.1 to 74.7)	<.001
HSCT in CR1	157	58.8 (50.5 to 66.1)		49.8 (41.7 to 57.4)		38.0 (30.4 to 45.6)

Abbreviations: CM, cytogenetic/molecular; CR1, first complete remission; HSCT, hematopoietic stem cell transplantation; MRD, measurable residual disease.

TABLE 3.

Univariable and Multivariable Survival Analysis of CM High-Risk Group

Univariable Survival Analysis CM High-Risk Group (n = 463)	No.	5 Year Overall Survival From End of Intensification I		5 Year Disease-Free Survival From End of Intensification I		5 Year Cumulative Incidence of Relapse From End of Intensification I
		HR (95% CI)	P	HR (95% CI)	P	HR (95% CI)	P
Age category
0-1 years	103	1.231 (0.87 to 1.74)	.242	1.342 (0.98 to 1.84)	.068	1.444 (1.03 to 2.03)	.034
2-10 years	168	1		1		1
≥11 years	192	1.287 (0.96 to 1.73)	.097	1.084 (0.82 to 1.43)	.564	0.877 (0.66 to 1.17)	.371
WBC count
<100,000/µL	365	1		1		1
≥100,000/µL	98	1.091 (0.80 to 1.49)	.583	1.126 (0.85 to 1.50)	.419	1.006 (0.73 to 1.39)	.971
Treatment
AAML0531 arm A	90	1		1		1
AAML0531 arm B	103	0.758 (0.52 to 1.11)	.156	0.715 (0.51 to 1.01)	.056	0.655 (0.46 to 0.93)	.018
AAML1031 arm A	100	0.942 (0.65 to 1.37)	.757	0.835 (0.59 to 1.19)	.323	0.721 (0.49 to 1.06)	.095
AAML1031 arm B	107	0.760 (0.52 to 1.11)	.160	0.809 (0.57 to 1.15)	.234	0.852 (0.60 to 1.21)	.365
AAML1031 arm C	63	0.512 (0.31 to 0.85)	.009	0.481 (0.30 to 0.76)	.002	0.359 (0.21 to 0.61)	<.001
Gemtuzumab ozogamicin
No	360	1		1		1
Yes	103	0.924 (0.68 to 1.26)	.617	0.895 (0.68 to 1.19)	.439	0.870 (0.65 to 1.17)	.357
MRD at end of induction I
Negative	277	1		1		1
Positive	143	1.356 (1.03 to 1.79)	.032	1.308 (1.01 to 1.70)	.045	1.361 (1.03 to 1.80)	.030
Unknown	43
Treatment following intensification I
Chemo in CR1	306	1		1		1
HSCT in CR1	157	0.670 (0.50 to 0.89)	.006	0.519 (0.40 to 0.68)	<.001	0.406 (0.31 to 0.54)	<.001

Multivariable Survival Analysis CM High-Risk Group (n = 420)a	No.	HR (95% CI)	P	HR (95% CI)	P	HR (95% CI)	P
Treatment
AAML0531 arm A	79	1		1		1
AAML0531 arm B	80	0.860 (0.57 to 1.31)	.480	0.776 (0.53 to 1.13)	.190	0.771 (0.52 to 1.15)	.206
AAML1031 arm A	98	1.002 (0.67 to 1.50)	.994	0.829 (0.57 to 1.21)	.332	0.722 (0.47 to 1.10)	.130
AAML1031 arm B	106	0.801 (0.54 to 1.19)	.275	0.848 (0.59 to 1.22)	.375	0.913 (0.62 to 1.34)	.641
AAML1031 arm C	57	0.581 (0.34 to 1.00)	.051	0.659 (0.40 to 1.10)	.110	0.559 (0.30 to 1.04)	.064
MRD at end of induction I
Negative	277	1		1		1
Positive	143	1.549 (1.16 to 2.08)	.003	1.648 (1.24 to 2.19)	.001	1.869 (1.36 to 2.57)	<.001
Treatment following intensification I
Chemo in CR1	275	1		1		1
HSCT in CR1	145	0.711 (0.52 to 0.98)	.036	0.518 (0.38 to 0.71)	<.001	0.386 (0.27 to 0.54)	<.001

Abbreviations: CM, cytogenetic/molecular; CR1, first complete remission; HR, hazard ratio; HSCT, hematopoietic stem cell transplantation; MRD, measurable residual disease.

^a Sample size based on number of patients with known MRD at end of induction I.

In the SR CM and EOInd1 MRD+ groups, 5-year OS was 39.6% (95% CI, 24.3 to 54.5) compared with 53.7% (95% CI, 35.1 to 69.0; P = .191) in patients receiving chemotherapy alone in CR1 (n = 39) and HSCT in CR1 (n = 33), respectively. DFS was 16.9% (95% CI, 4.3 to 36.7) compared with 50.9% (95% CI, 32.7 to 66.5; P = .032), and CIR was 74.3% (95% CI, 47.2 to 88.9) compared with 37.0 (95% CI, 20.5 to 53.5; P = .025; Fig 3; Table 2). Multivariable analysis was not performed because chemotherapy versus HSCT was the only variable associated with significant differences in outcome in univariable analysis (Data Supplement, Table S5).

In Horan et al,¹ we observed no differences between chemotherapy alone versus HSCT in CR1 on the basis of the previous historical definition of high-risk group (monosomy 7, monosomy 5, and deletion 5q). We performed the same analysis within our current study cohort. For patients with monosomy 7, monosomy 5, or deletion 5q, 5-year OS, DFS, and CIR were not significantly different in patients receiving chemotherapy alone versus HSCT in CR1 (Data Supplement, Table S6).

Subgroup Analyses

We performed several subgroup analyses. We examined the relative effect of HSCT in CR1 separately in patients with FLT3-ITD and in patients with all other forms of high-risk CM alterations (Data Supplement, Table S7). OS in patients with FLT3-ITD receiving chemotherapy alone in CR1 (n = 78) was 50.4% (95% CI, 38.5 to 61.1) and in patients receiving HSCT in CR1 (n = 92) was 65.4% (95% CI, 54.5 to 74.3; P = .056). DFS was 30.9% (95% CI, 18.8 to 43.8) and 53.7% (95% CI, 42.9 to 63.4), respectively (P = .001). CIR was 59.5% (95% CI, 44.5 to 71.6) and 35.4% (95% CI, 25.6 to 45.3), respectively (P = .001). OS in patients receiving chemotherapy alone in CR1 for non–FLT3-ITD high-risk CM alterations (n = 227) was 41.6% (95% CI, 35.0 to 48.1) and in patients receiving HSCT in CR1 (n = 64) was 48.3% (95% CI, 35.4 to 60.1; P = .454). DFS was 24.9% (95% CI, 19.0 to 31.2) and 43.3% (95% CI, 30.9 to 55.0), respectively (P = .017). CIR was 71.8% (95% CI, 64.8 to 77.6) and 42.5% (95% CI, 30.1 to 54.4), respectively (P < .001).

We assessed the impact of HSCT for patients with high-risk CM alterations in the AAML0531 and AAML1031 trials separately (Data Supplement, Table S8). OS in patients enrolled on AAML0531 receiving chemotherapy alone in CR1 (N = 126) was 41.7% (95% CI, 32.8 to 50.3) and in patients receiving HSCT in CR1 (n = 67) was 51.2% (95% CI, 38.5 to 62.5; P = .138). DFS was 23.7% (95% CI, 16.4 to 31.7) and 43.1% (95% CI, 31.1 to 54.5), respectively (P = .002). CIR was 71.3% (95% CI, 62.1 to 78.7) and 44.9% (95% CI, 32.6 to 56.4), respectively (P < .001). OS in patients enrolled on AAML1031 receiving chemotherapy alone in CR1 (N = 180) was 45.1% (95% CI, 37.5 to 52.5) and in patients receiving HSCT in CR1 (n = 90) was 64.9% (95% CI, 54.0 to 73.9; P = .015). DFS was 28.3% (95% CI, 20.7 to 36.3) and 55.0% (95% CI, 44.0 to 64.7), respectively (P < .001). CIR was 67.3% (95% CI, 58.3 to 74.8) and 32.8% (95% CI, 23.2 to 42.7), respectively (P < .001). We also performed multivariable analysis for AAML0531 and AAML1031 trials separately. For both AAML0531 and AAML1031, HSCT in CR1 was associated with improved DFS and CIR compared with chemotherapy alone in CR1 (Data Supplement, Tables S9 and S10).

Finally, we assessed the influence of EOInd1 MRD status on transplant outcomes in the high-risk CM group (Data Supplement, Table S11). Five-year OS in the MRD– (n = 72) and MRD+ (n = 73) groups was 63.1% (95% CI, 50.6 to 73.2) and 50.1% (95% CI, 37.8 to 61.2), respectively (P = .139). DFS was 56.6% (95% CI, 44.3 to 67.1) and 37.3% (95% CI, 26.2 to 48.5), respectively (P = .018), and CIR was 28.1% (95% CI, 18.2 to 38.9) and 52.9% (95% CI, 40.5 to 63.8), respectively (P = .001).

DISCUSSION

The results of our analysis indicate that pediatric patients with high-risk AML defined by contemporary criteria benefit from HSCT in CR1. This runs counter to the results of our previous studies, which examined the impact of HSCT in a much smaller high-risk group comprised primarily of patients with monosomy 7, monosomy 5, or deletion 5q, demonstrating no benefit. That research indicated instead that it was the patients with SR disease who derived benefit from HSCT.^1,2 A plausible explanation for these divergent results is that with the changes made to the risk classification schema since the early 2000s, approximately half the original SR group has been reallocated to the high-risk group.

Recent studies have investigated the impact of MRD on HSCT outcomes in pediatric AML.^20-22 However, our study is the first to incorporate the recent expansion of high-risk alterations uncovered by high-throughput sequencing^6-14 to investigate the impact of HSCT versus chemotherapy alone in CR1.

To perform direct comparisons between chemotherapy and HSCT following the prescribed three courses of chemotherapy that preceded HSCT, we restricted our analysis to patients who achieved morphologic CR and received three courses of pretransplant chemotherapy. Our exclusion of patients who do not achieve a CR omits patients who are unlikely to benefit from HSCT, since achieving an adequate antileukemia response before HSCT is a key determinant of outcome.¹⁶ Nevertheless, our results should be interpreted within this context and cannot be generalized to cases of refractory disease, where additional courses of chemotherapy or rationale targeted therapies (rather than HSCT without CR) should be considered.

Despite limitations associated with cross-study reports, separate analyses of the two trials revealed similar results, with the notable exception of no significant difference in OS for patients enrolled on AAML0531. There are a few potential reasons for this observation. AAML0531 is the older of the two trials, enrolling patients from 2006 to 2010. Therefore, less advanced HSCT-related supportive care measures may have affected OS. Similarly, more toxic conditioning regimens used for AAML0531 patients may have contributed to increased transplant-related mortality.

The trials included in our original cross-study analysis prospectively allocated patients to HSCT on the basis of the availability of an HLA-matched related donor. The use of donor availability to direct assignment is often referred to as biologic assignment or biologic randomization. Although this approach is not as scientifically rigorous as a true randomization, it continues to be used in trials comparing HSCT with nontransplant therapies, where randomization may deter participation or pose ethical challenges. A strength of this approach is that it allows intention to treat analysis on the basis of donor availability, often referred to as donor versus no donor analyses. A limitation of our current analysis is that it was not feasible to do an intention to treat analysis. The allocation of patients with high-risk disease on both trials to HSCT using the best donor available and the retrospective reclassification of risk precluded this.

Despite this significant limitation, we believe that the consistent association of HSCT with improved outcomes across multiple analyses in our study strongly suggests that pediatric patients with contemporarily defined high-risk AML benefit from HSCT in CR1. Consolidative transplantation was associated with improvement in outcomes in patients in the high-risk CM group, SR CM with EOInd1 MRD+ group, the subset of patients enrolled on the AAML0531 trial as well as the AAML1031 trial, the subset of patients with FLT3-ITD, and the subset of patients with other high-risk CM alterations, with the exception of chromosome 7 or 5 loss. Only 12 of 254 patients within the SR CM with EOInd1 MRD− group underwent HSCT in CR1, precluding our ability to draw conclusions for this group. Finally, in multivariable analysis for the high-risk CM group (the one group large enough for multivariable analysis), HSCT remained strongly associated with improved outcomes.

DATA SHARING STATEMENT

A data sharing statement provided by the authors is available with this article at DOI https://doi.org/10.1200/JCO-24-01841. The data generated for this study have been deposited in dbGaP under phs000465.v21.p8 (https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000465.v21.p8) and is available through controlled-access, as part of the NIH Genomic Data Sharing Policy to ensure that all approved investigators and institutions abide by the NIH Genomic Data User Code of Conduct, the terms of the Data Use Certification, and the Security Best Practices for Controlled Access Data (for more details, https://grants.nih.gov/grants/guide/notice-files/NOT-OD-14-124.html).

AUTHOR CONTRIBUTIONS

Conception and design: Benjamin J. Huang, Todd A. Alonzo, E. Anders Kolb, Soheil Meshinchi, Joseph H. Chewning, William G. Woods, John T. Horan

Provision of study materials or patients: Betsy Hirsch, Richard Aplenc, Alan S. Gamis, Gordana Raca

Collection and assembly of data: Todd A. Alonzo, Adam J. Lamble, Rhonda E. Ries, Weijie Wang, Betsy Hirsch, Gordana Raca, Richard Aplenc, E. Anders Kolb, Michael R. Loken

Data analysis and interpretation: Benjamin J. Huang, Lauren K. Meyer, Todd A. Alonzo, Yi-Cheng Wang, Adam J. Lamble, Weijie Wang, Xiaotu Ma, Alan S. Gamis, Richard Aplenc, E. Anders Kolb, Todd M. Cooper, Katherine Tarlock, Soheil Meshinchi, Joseph H. Chewning, John T. Horan

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Hematopoietic Stem Cell Transplantation Outcomes for High-Risk AML: A Report From the Children's Oncology Group

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).