Validation of the 2017 European LeukemiaNet Classification for Acute Myeloid Leukemia With NPM1 and FLT3-Internal Tandem Duplication Genotypes

Abstract

BACKGROUND:

The revised 2017 European LeukemiaNet (ELN) classification (ELN-2017) of acute myeloid leukemia (AML) divides patients into 3 prognostic risk categories, with additional factors such as the fms-like tyrosine kinase 3 (FLT3)-internal tandem duplication (ITD) allele ratio (AR) considered for risk stratification. To the best of the authors’ knowledge, the prognostic usefulness of ELN-2017 in comparison with ELN-2010 in younger patients with AML has not been validated to date.

METHODS:

The authors performed a retrospective study on patients aged <60 years who received idarubicin plus cytarabine (IA)-based induction chemotherapy for newly diagnosed AML.

RESULTS:

According to ELN-2017 criteria, the number of patients in the favorable (Fav), intermediate (Int), and adverse (Adv) risk categories was 192 patients (27%), 331 patients (46%), and 192 patients (27%), respectively. Overall survival probabilities at 5 years in the Fav, Int, and Adv groups were 57%, 37%, and 18%, respectively. In comparison, the 5-year overall survival probabilities in the Fav (169 patients), intermediate (IR)-1 (80 patients), IR-2 (306 patients), and Adv (160 patients) ELN-2010 categories were 59%, 32%, 40%, and 14%, respectively. Although ELN-2010 historically distinguishes prognosis into IR-1 and IR-2 categories in younger patients, this difference was nullified in the current study cohort. When comparing patients with a low FLT3-ITD AR with those with a high FLT3-ITD AR, no significant differences in survival were noted among patients with nucleophosmin 1 (NPM1)-mutated AML (P=.28) or wild-type NPM1 (P=.35), and in those treated with IA alone (P=.79) or those treated with IA and a FLT3 inhibitor (P=.10).

CONCLUSIONS:

The ELN-2017 more accurately distinguishes prognosis in patients with newly diagnosed AML. The lack of prognostic significance for the FLT3-ITD AR needs further evaluation in different treatment settings.

INTRODUCTION

Acute myeloid leukemia (AML) is a collection of molecularly and clinically heterogeneous disorders characterized by a myriad of cytogenetic and mutational abnormalities of varied prognostic significance.¹ Since the development of early cytogenetic-based risk classifications and their refinements,^2–4 an improved understanding of the molecular pathogenesis of AML has led to the improved accuracy of risk classification by combining molecular and cytogenetic data. The European LeukemiaNet (ELN) first recommended a risk classification system in 2010 (ELN-2010) in patients with AML that integrated cytogenetic and molecular data.⁵ Since then, further advances in our understanding of the genomic landscape and the application of improved assays for genetic testing have led to the development of prognostic models incorporating molecular data with the aim of improving AML risk stratification.^6–8

In the original ELN-2010 classification, 4 categories were recognized: favorable risk (FR), intermediate I risk (IR-1), intermediate II risk (IR-2), and adverse risk (AR), with IR-1 and IR-2 based on molecular as well as cytogenetic features. Subsequent validation studies demonstrated that the 2 IR groups were distinguishable only in younger patients but not in the elderly population.^9,10 This led to a proposal for a more simplified 3-group genetic risk stratification (favorable [Fav], intermediate [Int], and adverse [Adv]) in the revised ELN-2017 classification.¹¹ Furthermore, accumulating data regarding the prognostic relevance of recurrent genetic alterations have required several additional changes in the updated ELN recommendations.^5,11 These changes reflect an improved understanding of the prognostic influence of mutations and co-occurring mutation interactions on outcomes.⁷ Important considerations in subgroup assignment in ELN-2017 include stratification of nucleophosmin 1 (NPM1) and biallelic CCAAT/enhancer-binding protein alpha (CEBPA) mutations into the Fav risk group irrespective of associated karyotype; the recognition of the possible impact of the allelic ratio of fms-like tyrosine kinase 3 (FLT3)-internal tandem duplication (ITD) in risk stratification; and the inclusion of runt-related transcription factor 1 (RUNX1), ASXL transcriptional regulator 1 (ASXL1), and TP53 mutations and monosomal karyotype in the Adv risk category.

To the best of our knowledge, since the publication of the ELN-2017 recommendations, few studies to date have attempted to validate them.^12,13 Furthermore, the validity of such prognostication when higher dose cytarabine regimens are used for induction in younger patients with AML has not been examined. To address these questions, we evaluated the prognostic usefulness of the ELN-2017 risk stratification in comparison with that of the ELN-2010 in younger patients with AML (those aged <60 years) who were treated with higher dose cytarabine-based regimens in the study institution over the past 20 years.

MATERIALS AND METHODS

Study Design

A retrospective chart review was performed to evaluate patients aged 18 to 60 years with newly diagnosed AML and available clinical and genetic data who were treated at the study institution from January 2000 to December 2015 on and off clinical trials. The study population included those who received idarubicin plus high-dose cytarabine (IA)-based induction chemotherapy. Patients were administered cytarabine at a dose of either 1.5 g/m² daily for 4 days by continuous infusion or 1 to 2 g/m² daily for 5 days combined with idarubicin with or without fludarabine, cladribine, or clofarabine. Once they had achieved complete remission (CR) or CR with incomplete platelet recovery (CRp), patients received consolidation therapy with high-dose cytarabine or proceeded to allogeneic hematopoietic stem cell transplantation (SCT) in first CR (CR1). All protocols were approved by the institutional review board, and written informed consent was obtained before study enrollment, in accordance with the Declaration of Helsinki. A specific chart review protocol to evaluate predictors of outcome in patients with AML was approved by the institutional review board.

Patient Population

A diagnosis of AML was made based on the presence of ≥20% blasts in the bone marrow or peripheral blood.¹⁴ Cytogenetic analysis was performed on the baseline bone marrow sample using standard methods. From January 2000 to September 2012, molecular testing was performed using polymerase chain reaction (PCR) amplification followed by Sanger sequencing to detect specific mutations in candidate genes.¹⁵ From September 2012 onward, mutation testing was performed within a 28-gene panel using next-generation sequencing as previously described.¹⁶ Patients with CEBPA mutations were excluded because data from the cloning of full-length CEBPA open reading frames to clarify whether the different mutations were on the same or separate alleles was not available. In essence, all the patients in the current study cohort had CEBPA-nonmutated AML. Chimeric gene transcript testing was performed by reverse transcriptase-mediated quantitative PCR.¹⁶ To measure the allelic ratio of FLT3-ITD, exons 14 and 15 of the FLT3 gene were amplified from DNA by PCR using a fluorescently labeled primer, and the products were analyzed by fragment analysis. When a mutation was detected, the ratio of the area under the peak of mutant over total (mutant [mut] + wild-type [wt]) FLT3 was reported for monitoring the mutant allele burden. Patients with a FLT3-ITD mutation were classified as having a low FLT3 allele ratio (FLT3^low) (<0.5) or a high FLT3 allele ratio (FLT3^high) (≥0.5) based on the mutant allele ratio. Patients were assigned based on guidelines published in ELN-2010⁹ into FR, IR-1, IR-2, and AR groups, and Fav, Int, Adv groups as per the ELN-2017 recommendations.¹¹

Statistical Analysis

Descriptive statistics were presented as medians and ranges for continuous data and as numbers and percentages for categorical data. The chi-square test and Mann-Whitney U test were used to compare categorical and continuous data, respectively. Overall survival (OS) was calculated from the time of the AML diagnosis to the date of death and censored at the time of last follow-up if the patient was alive. Survival curves were established using the Kaplan-Meier method and were compared using the log-rank test. Variables with P ≤.05 were chosen for the multivariate Cox regression model. Covariates included baseline hematological parameters, antecedent hematological disorder, age, ELN categories, and molecular mutations. Cox regression analyses were fitted with baseline prognostic factors including cytogenetic and molecular mutation status, with receipt of allogeneic SCT as a time-dependent covariate. Hazard ratios (HRs) were denoted with 95% confidence intervals (95% CIs). Statistical significance was determined at a 2-tailed P value of .05.

RESULTS

Overall, a total of 715 patients with a median age of 51 years (range, 18–60 years) were included in the study, including 192 patients (27%), 331 patients (46%), and 192 patients (27%) in the Fav, Int, and Adv risk categories according to the ELN-2017 classification. Their characteristics are outlined in Table 1. There were no significant differences noted with regard to the age distribution among the 3 categories. The white blood cell count and lactate dehydrogenase levels were found to be significantly higher in the Fav group compared with the Int and Adv cohorts, and the platelet count was significantly lower in the Fav group compared with the Int and Adv cohorts. Of the patients in the Fav, Int, and Adv groups, 44 patients (23%), 94 patients (28%), and 71 patients (37%), respectively, underwent an allogeneic SCT (P<.001).

TABLE 1.

Patient Characteristics According to the ELN-2017 Classification

Characteristics	Overall N = 715	Favorable N = 192	Intermediate N = 331	Adverse N = 192	P
Age >50 y, no. (%)	373 (52)	97 (51)	172 (52)	107 (56)	.75
Median WBC (range), K/µL	6.5 (0.5-378.4)	11.7 (0.6-378.4)	5.6 (0.6-250)	5.3 (0.5-339.5)	<.001
Median platelet count (range), K/µL	40 (1-708)	31 (1-431)	41 (2-708)	47 (2-656)	<.001
Median LDH (range), IU/L	815(200-20527)	975 (315-11064)	679 (200-12489)	815 (231-20527)	<.001
Median BM blast percentage (range), %	52 (0-99)	57 (2-96)	55 (0-99)	45 (2-98)	.02
ECOG PS ≥2, no. (%)	141 (19.7)	18 (9.4)	89 (27)	34 (18)	<.001
AHD, no. (%)	40 (5.6)	3 (1.6)	20 (6)	17 (9)	.003

Abbreviations: AHD, antecedent hematologic disorder; BM, bone marrow; ECOG PS, Eastern Cooperative Oncology Group performance status; ELN-2017, 2017 European LeukemiaNet classification; LDH, lactate dehydrogenase; WBC, white blood cell count.

Risk Stratification According to the ELN-2010 and ELN-2017 Recommendations

In the ELN-2010 classification, the FR, IR-1, IR-2, and AR groups consisted of 169 patients (23.6%), 80 patients (11.2%), 306 patients (42.7%), and 160 patients (22.3%), respectively, indicating that many patients were recategorized from the IR-1 and IR-2 groups into the Fav and Adv categories using the ELN-2017 criteria. Distributions of genetic abnormalities among patients according to the ELN-2010 and ELN-2017 systems are shown in Table 2 and Supporting Table 1, respectively.

TABLE 2.

Distribution of Genetic Abnormalities According to the ELN-2017 Classification System

Risk Category	Genetic Abnormality	No. (%)
Favorable	t(8;21)(q22;q22)	24 (12.5)
	inv(16)(p13.1q22)/t(16;16)(p13.1q22)	45 (23.4)
	NPM1mutFLT3low/NPM1mutFLT3wt	123 (64.1)
	Total	192
Intermediate	NPM1mutFLT3high	24 (7.3)
	NPM1wtFLT3low/NPM1mutFLT3wt	125 (37.7)
	t(9;11)(p21.1;q23.3)	10 (3.0)
	Cytogenetic abnormalities not classified as favorable or adverse	172 (52)
	Total	331
Adverse	t(6;9)(p23;q34)	8 (4.2)
	t(v;11)(v;q23.3)	11 (5.7)
	t(9;22)(q34.1;q11.2)	1 (0.5)
	Monosomy 7	9 (4.7)
	Complex karyotypea	124 (64.6)
	Inv (3) or t(3;3)	8 (4.2)
	Monosomal karyotype	5 (2.6)
	NPM1wt FLT3high	17 (8.9)
	RUNX1mut/ASXL1mut/TP53mut	9 (4.7)
	Total	192

Abbreviations: ELN-2017, 2017 European LeukemiaNet classification; FLT3, fms-like tyrosine kinase 3; high, high allele ratio; low, low allele ratio; mut, mutant; NPM1, nucleophosmin 1; wt, wild type.

^aAll −5/5q- and 17p deletion/del(17) cases occurred as a part of a complex karyotype.

The ELN-2017 Fav group (192 patients) consisted of 168 patients with FR, 5 patients with IR-1, and 19 patients with IR-2 according to the ELN-2010 criteria (Fig. 1A). Of the patients recategorized into the Fav risk group, 5 had NPM1^mut FLT3^low (normal karyotype), 3 patients had NPM1^mut FLT3^low (abnormal, nonadverse karyotype), and 16 patients had NPM1^mut FLT3^wt (abnormal, nonadverse karyotype). The ELN-2017 Int group (331 patients) consisted of 1 patient in the FR group, 51 patients in the IR-1 group, and 279 patients in the IR-2 group according to the ELN-2010 criteria (Fig. 1B). Of the patients recategorized into the Int risk group, the lone FR patient had NPM1^mut FLT3^wt (abnormal, nonadverse karyotype). Of the 51 patients with IR-1 recategorized as Int, 23 had NPM1^mut FLT3^high (normal karyotype) and 28 had NPM1^wt FLT3^wt or NPM1^wt FLT3^low (normal karyotype). Of the 279 IR-2 patients reclassified into the Int category, 10 patients had t(9;11)(p21.1;q23.3) and 269 patients had cytogenetic abnormalities not classified as Fav or Adv. The ELN-2017 Adv group (192 patients) consisted of 24 patients with IR-1, 8 patients with IR-2, and 160 patients with AR according to the ELN-2010 criteria (Fig. 1C). Of the 24 patients with IR-1 who were recategorized into Adv, 17 had NPM1^wt FLT3^high (normal karyotype) and 7 had ASXL1^mut or RUNX1^mut (normal karyotype). Of the 8 patients with IR-2 recategorized into Adv, 5 patients had a monosomal karyotype, 1 patient had t(9;22), and 2 patients had ASXL1^mut or RUNX1^mut.

Figure 1.

Changes in the risk categories between the 2010 European LeukemiaNet (ELN-2010) classification system and the ELN-2017 classification system with the ELN-2017 system.

Prognostic Analysis According to the ELN-2017 Classification

CR/CRp was achieved in 590 patients (83%) (see Supporting Table 2). The CR/CRp rates were higher in the Fav risk group (96%) and Int risk group (83%) compared with patients in the Adv risk group (69%). Using the ELN-2010 criteria, CR/CRp rates in the FR, IR-1, IR-2, and AR groups were 95%, 86%, 84%, and 66%, respectively (see Supporting Table 2). Although the ELN-2010 discriminated between groups based on CR/CRp rates overall (P<.001), it did not distinguish between the IR-1 and IR-2 risk groups (P=.50).

The median follow-up among survivors for the overall study population was 16.4 months (range, 1.0–140.9 months). The median OS in the Fav, Int, and Adv ELN-2017 groups was 79.1 months, 26.9 months, and 11.2 months, respectively (P<.001) (Fig. 2A).

Figure 2.

Overall survival according to the 2017 European LeukemiaNet (ELN-2017) classification system and ELN-2010 risk categories (A) overall, irrespective of transplantation status and (B) when censoring for allogeneic stem cell transplantation.

In comparison, the median OS in the FR, IR-1, IR-2, and AR ELN-2010 groups was 79.7 months, 28.0 months, 30.1 months, and 10.2 months, respectively (IR-1 vs IR-2; P=.12) (Fig. 2A). A similar pattern was observed when censoring for transplantation (Fig. 2B), with an inferior survival noted among patients in the IR-1 group when compared with the IR-2 group, although it was statistically nonsignificant (13.7 months vs 26.2 months, respectively; P=.15). To put the survival impact of transplantation into perspective, the 3-year OS probabilities in the IR-1 group with and without censoring for allogeneic SCT were 25% and 42%, respectively.

We then compared survival between the FR and IR-2 groups within the ELN-2017 Fav group and found no difference with regard to OS (P=.72) (Fig. 3). In the Int group, survival among patients in the IR-2 group was superior to that of patients in the IR-1 group (P=.03) (Fig. 4A), with a trend toward superior survival on censoring for transplantation (P=.09) (Fig. 4B). When comparing survival between the IR-1 group of patients who were recategorized into the Fav group and those IR-1 patients assigned to the Int group, we observed a statistically improved survival in the former (P=.02) (see Supporting Fig. S1). Within the ELN-2017 Adv group, OS was significantly different between the IR-1 and AR groups of the ELN-2010 (P<.001) (Fig. 5). Further analyses could not be performed to assess the relative prognosis of IR-1 patients in the Fav group, FR patients in the Int group, and IR-2 patients in the Adv group due to small sample sizes.

Figure 3.

Overall survival according to 2010 European LeukemiaNet (ELN-2010) classification subgroups within the ELN-2017 favorable risk category (Fav). IR-2 indicates intermediate II risk.

Figure 4.

Overall survival according to 2010 European LeukemiaNet (ELN-2010) classification subgroups within the ELN-2017 intermediate risk category (A) overall, irrespective of transplantation status and (B) when censoring for allogeneic stem cell transplantation. IR-1 indicates intermediate I risk; IR-2, intermediate II risk.

Figure 5.

Overall survival according to 2010 European LeukemiaNet (ELN-2010) classification subgroups within the ELN-2017 adverse risk category (Adv). IR-1 indicates intermediate I risk.

Finally, we compared the prognostic impacts of each risk category based on the ELN-2017 (see Supporting Table 3) and ELN-2010 (see Supporting Table 4) classifications, respectively. Although both risk categories clearly discriminated prognoses between Fav, Int, and Adv risk groups, the ELN-2010 did not distinguish prognostically between the IR-1 and IR-2 groups in a subanalysis.

Time-Dependent Analysis of Allogeneic SCT

We performed an extended Cox regression analysis with allogeneic SCT in CR1 as a time-dependent covariate. The impact differed among patients assigned to the 3 risk groups defined by the ELN-2017 recommendations, with the results indicating that allogeneic SCT had no impact on OS in patients with Fav AML (HR, 1.13; 95% CI, 0.59–2.29) (see Supporting Table 5), but did have a beneficial impact in patients with IR AML (HR, 0.46; 95% CI, 0.29–0.71) (see Supporting Table 6) and high-risk AML (HR, 0.45; 95% CI, 0.30–0.69) (see Supporting Table 7).

Prognosis According to NPM1 and FLT3-ITD Mutation Status

We categorized patients without adverse-risk genetic lesions into 4 groups as proposed in ELN-2017 based on NPM1 and FLT3 mutation status and the FLT3 allele ratio: 1) NPM1^mutFLT3^low or NPM1^mutFLT3^wt (123 patients); 2) NPM1^mutFLT3^high (24 patients); 3) NPM1^wtFLT3^low or NPM1^wtFLT3^wt (125 patients); and 4) NPM1^wtFLT3^high (17 patients). The median OS for groups 1, 2, 3, and 4 was 79.7 months, 17.2 months, 25.7 months, and 42.5 months, respectively (P=.02) (Fig. 6). Although groups 2 and 3 were associated with a significantly inferior survival when compared with group 1, NPM1^wtFLT3^high genotype status (group 4) did not appear to have an inferior prognosis compared with group 1 (see Supporting Table 8). The high transplantation rate of 59% most likely contributed to the particularly favorable outcome noted in patients in group 4; in comparison, transplantation rates in groups 1, 2, and 3 were 31% (vs group 4; P=.02), 23% (vs group 4; P=.002), and 50% (vs group 4; P=.57). Due to the very few noncensored observations available in group 4 for a desirable analysis, we did not undertake an allogeneic SCT-censored survival analysis between the 4 groups.

Figure 6.

Overall survival according to the nucleophosmin 1 (NPM1) and fms-like tyrosine kinase 3 (FLT3) genotypes as per the 2017 European LeukemiaNet (ELN-2017) classification system. FLT3^high indicates high FLT3 high allele ratio; FLT3^low, low FLT3 allele ratio; mNPM1, NPM1-mutated acute myeloid leukemia; wtNPM1, wild-type NPM1.

We also assessed whether the FLT3-ITD allelic ratio affected prognosis in the current study cohort. Of the 116 patients with a FLT3-ITD mutation, 33 patients (28%) were treated with IA along with a FLT3 inhibitor. We compared the prognosis of patients according to allele status, and did not observe significant differences in survival between patients in the FLT3^high and FLT3^low groups in the overall population (see Supporting Fig. S2A), among patients with NPM1^mut (see Supporting Fig. S2B), among patients with NPM1^wt (see Supporting Fig. S2C), in those treated with IA alone without a FLT3 inhibitor (see Supporting Fig. S2D), and in patients receiving IA combined with a FLT3 inhibitor (see Supporting Fig. S2E).

DISCUSSION

In the current study, we stratified patients according to the ELN-2010 and ELN-2017 classifications. Both classifications clearly distinguished prognoses between favorable, intermediate, and adverse risk categories. However, ELN-2010 did not distinguish prognosis between IR-1 and IR-2 risk groups. In fact, the distinction between IR-1 and IR-2 originally was based on genetic characteristics and not on prognostic stratification in any particular AML cohort. Consistent with findings from 2 previously published ELN-2010 validation studies,^9,10 the IR-2 group in the current study cohort was associated with a longer survival compared with the IR-1 group, albeit one that was statistically nonsignificant, when cases were censored for transplantation. However, this trend was lost when taking transplantation into account, with an improved survival noted in patients in the IR-1 group. This significant impact of transplantation on prognostic stratification in the IR risk category corroborates prior data from Rollig et al,¹⁰ who reported that no significant differences between the FR, IR-1, and IR-2 groups were noted when analyzing only those patients receiving allogeneic SCT in CR1.

Because the ELN-2017 guidelines recommend including patients with NPM1 mutations under the Fav risk category regardless of their underlying karyotype, we analyzed survival between the FR and IR-2 subcategories within the Fav group to evaluate whether their prognoses were comparable. The results indicated that there was no difference in OS between the 2 groups, thereby supporting the reclassification of NPM1-mutated patients into the Fav risk group irrespective of the associated karyotype. It must be pointed out that the patient numbers in the current study were small and further investigation in a larger number of patients still is needed. Although the current study data were broadly supportive of the 3-category ELN system, we observed within-group prognostic heterogeneity such as between the IR-1 and IR-2 subcategories in the ELN-2017 Int group. However, patients in the IR-1 group within the Fav group still had better outcomes than patients in the IR-1 group within the Int category, thereby justifying categorization based on NPM1 mutation status irrespective of underlying low FLT3-ITD allele burden status.

We also noted a significant survival difference between patients in the IR-1 (NPM1^wtFLT3^high) and AR subcategories in the ELN-2017 Adv group. Contrary to expectations, the NPM1^wtFLT3^high subgroup had highly favorable outcomes, with a survival that was comparable to that of the patients in the NPM1^mutFLT3^low or NPM1^mutFLT3^wt Fav subgroup, indicating that the allele ratio did not impact prognosis in the current study cohort. However, these data must be interpreted with caution given the high rate of transplantation in the NPM1^wtFLT3^high subgroup. Corroborating with this, allogeneic SCT in CR1 previously has been shown to outweigh the negative impact of a high allele ratio on survival.¹⁷

The most controversial issue in the cohort in the current study was the stratification into genetic categories based on the allele ratio. A key change in ELN-2017 is the reclassification of NPM1^mutFLT3^low as Fav in addition to NPM1^mut FLT3^wt, and the assignment of NPM1^wt FLT3^high under the high-risk category. These proposals were made based on recent data suggesting that NPM1-mutant status conveyed a favorable prognosis only in the absence of or with a low level of FLT3-ITD mutation and that wild-type NPM1 and FLT3-ITD with a high (≥0.5) allelic ratio (FLT3-ITD^high) was associated with inferior survival outcomes.^17–20 In contrast, results from the current analysis suggest that survival is not differentially impacted by the FLT3-ITD allele ratio as was the observation made in a recently published validation study by Harada et al.¹³ Moreover, we did not identify a survival difference based on the FLT3-ITD allele ratio despite excluding those patients treated with a FLT3 inhibitor. We speculate that this survival difference possibly was nullified by the use of high doses of cytarabine. The treatment regimens in the study institution include cytarabine administered at doses of >1 g/m², which is significantly higher than the traditional dosing of 100 to 200 mg/m² used in other institutions^17–20 and standard practice.²¹ To the best of our knowledge, the survival benefit of using higher doses of cytarabine in patients with AML with a high FLT3 mutation burden has not been explored to date. In a previously reported clinical experience from our institution, there was no survival benefit noted when using higher doses of cytarabine during induction in any of the 4 genetic subsets defined by NPM1 and FLT3-ITD genotype in patients newly diagnosed with AML.²² However, the FLT3-ITD allele ratio was not considered in the analysis. Although the noninferior outcomes may reflect the high rate of allogeneic SCT in the NPM1^wt and FLT3^high cohort and are influenced by the small sample size estimates, the differential impact of high-dose cytarabine on postinduction outcome based on the FLT3-ITD allelic ratio requires further investigation.

In the RATIFY trial, which evaluated intensive induction and consolidation chemotherapy plus midostaurin or placebo, the use of a FLT3 inhibitor improved survival in both of the midostaurin-treated arms irrespective of the allele ratio (defined by a cutoff value of 0.7).²³ In a post hoc analysis of the NPM1/FLT3-ITD genotypes as defined by the ELN-2017 from randomized patients treated within the trial, the beneficial effect of midostaurin on OS was most pronounced in the NPM1^wt and FLT3^high group.²⁴ Based on these data, the proposed ELN-2017 recommendations may change in the future, with the routine use of FLT3 inhibitors as the standard of care in patients with FLT3-mutated AML.

The decision to perform allogeneic SCT is a risk-adapted approach that depends on the assessment of the risk-benefit ratio based on cytogenetic and molecular genetic features along with patient, donor, and transplantation factors.²⁵ A subject of frequent debate is the value of allogeneic SCT as a postremission therapy among patients with intermediate-risk AML.²⁶ The time-dependent analysis of allogeneic SCT indicated that patients with favorable AML should not receive allogeneic SCT as postremission therapy in CR1, whereas patients with intermediate-risk and high-risk AML would benefit from allogeneic SCT in CR1.

A few limitations to the current study must be acknowledged. In addition to its retrospective study design and single-institution validation, outcomes from the current study may not be representative for patients treated with the standard 7 + 3 induction therapy using intermediate doses of cytarabine. The superior outcomes of patients in the IR-1 group (reclassified as Adv in ELN-2017) compared with AR patients in ELN-2010, and the equivalent OS outcomes between patients in the IR-2 group (reclassified as Fav in ELN-2017) and FR patients in ELN-2010 must be interpreted with caution considering the small size of the patient sample. Finally, patients with CEBPA mutations were not included, and CEBPA mutations are an integral part of the ELN-2017 classification. Nevertheless, the current study was not intended to be a comprehensive and exhaustive validation of the ELN-2017 classification but rather to specifically address the recategorization based on NPM1 and FLT3 genotype status, and the FLT3 allele ratio. Finally, there was the inclusion of a small percentage of patients who had a very short follow-up period, which may have distorted the study results.

In the current study, the ELN-2017 classification system was able to clearly distinguish long-term prognosis in adult patients with newly diagnosed AML compared with the ELN-2010 classification. An important change to the updated ELN guidelines is the inclusion of quantitative information regarding FLT3-ITD to improve risk stratification. However, risk stratification according to FLT3-ITD allele ratios is controversial and requires further evaluation in different treatment settings, such as the use of FLT3 inhibitors and allogeneic SCT.

The 2017 European LeukemiaNet (ELN-2017) classification system more accurately distinguishes prognosis compared with the ELN-2010 in patients with de novo acute myeloid leukemia. The prognostic significance of the fms-like tyrosine kinase 3 (FLT3)-internal tandem duplication allele ratio is controversial and needs further evaluation in different treatment settings.