REGULAR ARTICLE European LeukemiaNet 2017 risk stratification for acute myeloid leukemia: validation in a risk-adapted protocol Alex Bataller,1,2 Ana Garrido,2,3 Francesca Guijarro,1 Guadalupe O~nate,2,3 Marina Diaz-Bey a,1 Montserrat Arnan,4 Mar Tormo,5 Susana Vives, Marıa Paz Queipo de Llano, Rosa Coll, David Gallardo, Ferran Vall-Llovera,9 Lourdes Escoda,10 6 7 8 8 Antonio Garcia-Guin ~on,11 Olga Salamero,12 Anto nia Sampol,13 Brayan M. Merchan,14 Joan Bargay,15 Sandra Castan ~o-Dıez,1 1 Daniel Esteban, Aina Oliver-Calde s, Andrea Rivero, Pablo Mozas, Mo 1 1 1 nica Lopez-Guerra, 16,17 Marta Pratcorona, 2,3 Lurdes Zamora,2,6 Dolors Costa,16,17 Maria Rozman,16 Josep F. Nomdede u,2,3 Dolors Colomer,16,17 Salut Brunet,2,3 Jorge Sierra,2,3 and Jordi Esteve,1,2 on behalf of the Grupo Cooperativo Para el Estudio y Tratamiento de las Leucemias Agudas y Mielodisplasias (CETLAM) 1 Hematology Department, Hospital Clınic de Barcelona, August Pi i Sunyer Biomedical Research Institute (IDIBAPS), University of Barcelona (UB), Barcelona, Spain; 2Josep Carreras Leukemia Research Institute, Barcelona, Spain; 3Hematology Department, Hospital de la Santa Creu i Sant Pau, Universitat Aut onoma de Barcelona (UAB), Barcelona, Spain; 4Hematology Department, Catalan Institute of Oncology (ICO)–Hospital Duran i Reynals, L’Hospitalet de Llobregat, Spain; 5Hematology Department, Hospital Clınico Universitario de Valencia, Valencia, Spain; 6Hematology Department, Catalan Institute of Oncology (ICO)–Hospital Germans Trias i Pujol, Badalona, Spain; 7Hematology Department, Hospital Virgen de la Victoria, Malaga, Spain; 8Hematology Department, Catalan Institute of Oncology (ICO)–Hospital Universitari Dr. Josep Trueta, Girona, Spain; 9 Hematology Department, Hospital Universitari M utua de Terrassa, Terrassa, Spain; 10Hematology Department, Catalan Institute of Oncology (ICO)–Hospital Universitari Joan XXIII, Tarragona, Spain; Hematology Department, Hospital Arnau de Vilanova, Lleida, Spain; 12Hematology Department, Hospital Universitari Vall d’Hebron, Barcelona, Spain; 11 13 Hematology Department, Hospital Universitari Son Espases, Palma de Mallorca, Spain; 14Hematology Department, Hospital del Mar, Barcelona, Spain; 15Hematology Department, Hospital Universitari Son Llatzer, Palma de Mallorca, Spain; 16Hematopathology Section, Hospital Clınic de Barcelona, August Pi i Sunyer Biomedical Research Institute (IDIBAPS), University of Barcelona (UB), Barcelona, Spain; and 17Centro de Investigaci on Biomedica en Red en Oncologıa, Madrid, Spain Key Points The 2017 European LeukemiaNet (ELN 2017) guidelines for the diagnosis and management of acute myeloid leukemia (AML) have become fundamental guidelines to assess the prognosis  The ELN 2017 and postremission therapy of patients. However, they have been retrospectively validated in classification has few studies with patients included in different treatment protocols. We analyzed 861 patients been validated in a included in the Cooperativo Para el Estudio y Tratamiento de las Leucemias Agudas y risk-adapted intensive Mielodisplasias-12 risk-adapted protocol, which indicates cytarabine-based consolidation for protocol, supporting its utility to predict patients allocated to the ELN 2017 favorable-risk group, whereas it recommends allogeneic outcome. stem cell transplantation (alloSCT) as a postremission strategy for the ELN 2017 intermediate- and adverse-risk groups. We retrospectively classified patients according to the ELN 2017,  Within the ELN 2017 with 327 (48%), 109 (16%), and 245 (36%) patients allocated to the favorable-, intermediate-, adverse group, there and adverse-risk group, respectively. The 2- and 5-year overall survival (OS) rates were 77% is a subset of patients (inv(3) and TP53 and 70% for favorable-risk patients, 52% and 46% for intermediate-risk patients, and 33% and abnormalities) with a 23% for adverse-risk patients, respectively. Furthermore, we identified a subgroup of patients particularly poor within the adverse group (inv(3)/t(3;3), complex karyotype, and/or TP53 mutation/17p abnor- prognosis. mality) with a particularly poor outcome, with a 2-year OS of 15%. Our study validates the ELN 2017 risk stratification in a large cohort of patients treated with an ELN-2017 risk-adapted protocol based on alloSCT after remission for nonfavorable ELN subgroups and identifies a genetic subset with a very poor outcome that warrants investigation of novel strategies. Submitted 21 June 2021; accepted 8 November 2021; prepublished online on Blood © 2022 by The American Society of Hematology. Licensed under Creative Advances First Edition 15 December 2021; published online 15 February 2022. DOI Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC- 10.1182/bloodadvances.2021005585. ND 4.0), permitting only noncommercial, nonderivative use with attribution. All other Requests for data sharing may be submitted to Alex Bataller (Alex Bataller). rights reserved. The full-text version of this article contains a data supplement. 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 1193 Introduction Patients and methods Patients and treatment Acute myeloid leukemia (AML) is a heterogeneous clonal dis- Newly diagnosed patients with de novo AML eligible for intensive ease of transformed hematopoietic precursors presenting with chemotherapy and treated according to the CETLAM-12 protocol recurrent genetic abnormalities. In recent years, the unraveling from January 2012 to December 2019 were included in the study of the AML genetic landscape has allowed a better correlation (n 5 861). The CETLAM-12 protocol is detailed in the supplemen- of the biological features of the disease with the prediction of tal Materials. Essentially, all patients received a common induction response to treatment and clinical outcome.1 Based on that, regimen based on a 3 1 7 scheme with idarubicin and cytarabine. expert working groups have published recommendations for the Afterward, patients received a risk-based postremission therapy diagnosis, response criteria, and prognostic stratification of according to 3 categories based on initial genetics (including cyto- AML.2-4 In 2010, a panel of experts proposed the first European genetics, NPM1, FLT3-ITD, and CEBPA mutational status), and LeukemiaNet (ELN) recommendations for the diagnosis and postconsolidation minimal/measurable residual disease (MRD). The management of AML. The ELN 2010 risk classification stratified favorable group was defined as in the ELN 2017 classification, patients into 4 risk categories (favorable, intermediate I, interme- including the subset of patients with NPM1-mutated and wild-type diate II, and adverse) according to cytogenetics and NPM1, or low-ratio FLT3-ITD, given the favorable outcome observed in our FLT3, and CEBPA mutational status. Several groups have previous protocol.10 In these patients, consolidation therapy con- undertaken the effort to validate the prognostic impact of this sisted of 3 courses of high-dose cytarabine (HDAC). In these classification, although differential prognosis of intermediate I patients, bone marrow MRD assessment with quantitative reverse and intermediate II categories could not be confirmed.5,6 transcriptase-polymerase chain reaction (qRT-PCR) was performed after each chemotherapy course and thereafter at 3-month intervals In 2017, a revised version of the ELN classification was published for at least 3 years after CR. If a confirmed molecular failure was evi- (ELN 2017).4 This updated risk stratification grouped patients into 3 denced (ie, persistence or MRD reappearance after consolidation), risk categories (favorable, intermediate, and adverse risk) and alloSCT was the recommended strategy without a predefined refined the prognostic value of specific genetic mutations. Thus, guideline for specific pretransplant salvage chemotherapy. For CEBPA mutation defined a favorable AML subset only in a biallelic patients allocated to the intermediate- and high-risk categories, status, the presence of RUNX1, ASXL1, and TP53 mutations con- alloSCT was the recommended strategy after at least 1 HDAC ferred an adverse risk, and risk of FLT3 internal tandem duplication course. In addition, all patients with persistent MRD after consolida- (FLT3-ITD) was further modulated by its allelic ratio and the interac- tion therapy were allocated to the high-risk category, and alloSCT tion with NPM1 mutation. was also recommended. Since 2017, patients with FLT3 mutations (either ITD or FLT3-tyrosine kinase domain,TKD) could receive mid- Although the ELN 2017 risk classification was mainly designed ostaurin through an early access program. using data from the published evidence of individual genetic entities contained in the classification, to date, few studies have The CETLAM-12 protocol was approved by the Spanish Medicines validated its whole prognostic value in a large series of patients. Agency as a postapproval study (EPA-OD, JSG-SMD-2014-01). A recent publication by Herold et al7 validated the ELN 2017 Patients provided their written informed consent in accordance with among 771 available patients from 2 treatment protocols, with a the Declaration of Helsinki, and all procedures were approved by broad age range, comparing the prognostic impact according to the ethics committee. the both ELN classifications. This work highlighted the predic- Cytogenetic and mutational analysis tive prognostic potential of this latest classification, although it proposed a refined ELN 2017 stratification with 2 additional Cytogenetic analysis including fluorescence in situ hybridization groups (namely, very favorable and very adverse subgroups). Of was performed in the laboratory of each treating center. Mutational note, allogeneic hematopoietic stem cell transplantation analysis of NPM1 and FLT3 (ITD and TKD) was performed on (alloSCT) in first complete remission (CR1) had only been per- DNA, using PCR-based methods.10-12 Testing for recurrent AML formed in a minority of patients (16% of patients achieving CR). mutations, including RUNX1, ASXL1, CEBPA, NPM1, FLT3, and Other studies have also validated the prognostic impact of the TP53, was done with next-generation sequencing (NGS) techni- ELN 2017 classification, although with a smaller number of ques, with a limit of detection of 5% variant allele frequency. patients of younger age.8,9 Recurrent gene rearrangements, including RUNX1-RUNXT1 and CBFb-MYH11, were analyzed by qRT-PCR, fluorescence in situ The Spanish AML cooperative group Cooperativo Para el Estu- hybridization, or NGS. Participating centers and NGS panels used dio y Tratamiento de las Leucemias Agudas y Mielodisplasias are detailed in the supplemental Material. (CETLAM) proposed an AML risk-adapted treatment protocol (CETLAM-12), defining risk categories that virtually overlapped Risk group and ELN category allocation with the ELN 2017 classification, and recommending alloSCT in According to cytogenetic and mutational findings, each patient with CR1 for intermediate- and high-risk patients. Herein we present available data were retrospectively allocated to the corresponding the validation of the ELN 2017 risk stratification in patients with ELN 2017 risk category (favorable, intermediate, and adverse) to AML included in the CETLAM-12, with a genetic risk-based pre- analyze the outcome according to this classification. Of note, established transplant policy, unraveling a genetically defined patients with wild-type NPM1 and FLT3-ITD $ 0.5 lacking NGS very-high-risk subset within the adverse risk ELN 2017 testing were assigned to the adverse-risk category, irrespective of category. the presence or absence of adverse risk-defining gene mutations 1194 BATALLER et al 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 such as RUNX1, ASXL1, or TP53. Ambiguous cases and the frontline midostaurin during chemotherapy because of the presence detailed risk allocation procedure are detailed in the supplemental of a FLT3 mutation. Material. Treatment outcome according to ELN 2017 category Statistical methods All patient included in this study received at least induction therapy Characteristics among ELN 2017 groups were compared using the according to the CETLAM-12 protocol. A detailed flowchart with x2 test and Fisher's exact test for categorical variables and using patient disposition following induction and postremission therapy is the Student t test and Mann-Whitney U test for continuous variables shown in Figure 2. with normal and non-normal distribution, respectively. Median follow- Among patients allocated to the ELN 2017 favorable risk group (n 5 up time was calculated with the Kaplan-Meier estimate of potential 327), 295 achieved CR1 (90%) after 1 (n 5 277) or 2 (n 5 18) follow-up.13 Overall survival (OS) was calculated from diagnosis to induction courses. Although not intended in CR1, 82 patients (25%) death. Event-free survival (EFS) was calculated from diagnosis to treatment failure (not achieving complete remission/partial remission received an alloSCT: after experiencing an overt hematologic relapse after induction), relapse, or death. OS and EFS were estimated with in 29 patients (second complete response [CR2] n 5 25; partial the Kaplan-Meier method, and groups were compared with the log- response/refractory disease, n 5 4), after presenting molecular rank test.14,15 For outcomes where a possible competing event relapse (n 5 40; 33 patients with NPM1 mutation), and because of exists, cumulative incidence was calculated using relapse as the pri- a protocol deviation (n 5 13). Overt hematologic relapses were mary event and death without relapse as a competing event. Cumu- mostly detected after completion of HDAC consolidation courses lative incidence function comparison was performed using Gray's (44; 70% of all relapses). After alloSCT, 13 patients relapsed, and test. In the univariate and multivariate analyses, logistic regression 14 additional patients died because of a non–relapse-related cause was used to estimate odds ratio, whereas Cox proportional hazards (NRM). The median follow-up time of the ELN 2017 favorable group regression was used to estimate hazard ratios. Post hoc P value was 38 months, with 234 patients in a CR status (72% of all ELN analyses were adjusted with the Bonferroni method. All statistical 2017 favorable patients) at the cutoff time and 57 of them after an analyses were performed with R statistics version 4.0.2 (R core alloSCT (24% of all patients with CR). Among patients who did not Team, R Foundation for Statistical Computing, Vienna, Austria). receive an alloSCT, 10 patients died because of treatment complica- tions. Overall, 170 patients remained in sustained CR after HDAC- based consolidation chemotherapy (Figure 2). Results Regarding patients allocated to the ELN 2017 intermediate-risk Patient characteristics group (n 5 109), 87 achieved CR1 (80%), 18 of them requiring This study included a total of 861 patients enrolled in the multicen- more than 1 chemotherapy course. Thirteen early relapses before tric CETLAM-12 treatment protocol, classified according to the ELN alloSCT were observed, with 7 patients achieving CR2 after salvage 2017 classification as favorable (n 5 327), intermediate (n 5 109) therapy and therefore proceeding to alloSCT. Overall, 80 patients and adverse risk (n 5 245), corresponding to 48%, 16%, and 36% (73.4%) received an alloSCT (CR1, n 5 69, 63%; CR2, n 5 7; of classifiable patients, respectively. One hundred eighty additional refractory disease, n 5 4), and 3 patients received an autologous patients (20.9% of all patients) could not be classified according to stem cell transplant. After alloSCT, 19 patients relapsed, and 10 the ELN 2017 because of the lack of risk-defining cytogenetic patients died because of NRM. The median follow-up time of the abnormalities and unavailable NGS testing. This nonclassifiable ELN 2017 intermediate group was 24 months, with 57 patients in a group is mostly composed by ELN 2017 intermediate- and adverse- CR status (52% of all ELN intermediate-risk patients) at the cutoff risk patients with AML, without either favorable-risk mutations or time and 53 after an alloSCT (93% of all patients with CR). ELN group-defining cytogenetic findings, in whom NGS testing was not performed and RUNX1, ASXL1, and TP53 mutational status Concerning patients of the ELN 2017 adverse-risk group (n 5 245), could not be assessed. As a consequence, these patients could not 173 achieved CR1 (71%), 38 requiring more than 1 chemotherapy be properly allocated into either the ELN 2017 intermediate- or course. After achieving CR1, 41 patients relapsed, and only 5 of adverse-risk group. The main characteristics of each group and the them achieved CR2 after salvage therapy (12%). Altogether, 150 whole cohort are summarized in Table 1, including the comparison (61.6%) proceeded to alloSCT (CR1, n 5 125, 51%; CR2, n 5 3; with the Medical Research Council (MRC) and ELN 2010 risk clas- refractory disease, n 5 22), and 2 patients received an autologous sifications.3,16 Furthermore, the distribution of the genetic groups stem cell transplant. After alloSCT, 53 patients relapsed and 34 died within the ELN 2017 risk classification is detailed in Figure 1. because of NRM. The median follow-up time of the ELN 2017 adverse group was 38 months, with 68 patients in a CR status The median age of patients in the ELN 2017 adverse risk category (28% of all ELN adverse-risk patients) at the cutoff time and 66 after was higher than that of patients from the remaining risk groups, an alloSCT (97% of all patients with CR). without achieving statistical significance. These patients also had a lower white blood cell count (WBC), as well as lower peripheral Patients of the ELN 2017 favorable-risk group exhibited a higher blood and bone marrow blast count, compared with the other risk CR rate (90%) compared with the intermediate and adverse groups categories (P , .01), and a trend to a lower hemoglobin level. (80% and 71%, respectively; P , .05). Furthermore, the ELN 2017 Intermediate-risk patients presented with a higher WBC and blast adverse group showed a higher proportion of induction refractori- count than the other risk categories (P , .01), largely attributable to ness than the other groups (19% in the adverse group vs 1% and the higher prevalence of FLT3-ITD in this cohort (66.9% in the 8% in the favorable and intermediate groups, respectively; P , intermediate-risk group vs 15.6% and 18% among favorable- and .001). The multivariate analysis confirmed the independent prognos- adverse- risk patients, respectively). Of note, 46 patients received tic value of this classification for the achievement of CR after 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 VALIDATION OF THE AML ELN 2017 RISK STRATIFICATION 1195 Table 1. Patient characteristics of the entire cohort and according to the ELN 2017 risk stratification ELN 2017 favorable ELN 2017 ELN 2017 not All patients (861) (327) intermediate (109) ELN 2017 adverse (245) classifiable (180) P* Age, years median 56 (17-76) 54 (17-76) 53 (20-75) 58 (18-71) 59 (17-71) .33 (range) Sex, male n (%) 449 (51) 164 (50) 48 (44) 123 (50) 107 (59) .43 WBC, median (range) 9.88 (0.1-530) 17.9 (0.6-361) 45.6 (0.8-384.2) 5.2 (0.4-313.6) 3.4 (0.1-530) ,.001 (/109/L) Hemoglobin, median 89 (26-159) 92 (34-150) 91 (40-153) 87 (38-159) 87 (26-147) .08 (range) (g/L) Platelet count, median 62 (2-742) 60 (5-453) 62 (10-488) 62 (6-465) 74 (2-742) .96 (range) (/109/L) PB blasts, median 30 (0-100) 37 (0-100) 61 (0-100) 20 (0-99) 11 (0-98) ,.001 (range) (%) BM blasts, median 62 (1-100) 66 (2-100) 79 (17-100) 57 (7-99) 50 (9-99) ,.001 (range) (%) NPM1 status NPM1 mutated 295 (34.3) 217 (66.4) 72 (66.1) 6 0 — NPM1 unknown 27 (3) 5 (1.6) 0 10 (9.2) 12 (6.7) — FLT3-ITD status FLT3-ITDlow 60 (7.3) 45 (13.8) 2 (1.8) 7 (2.9) 6 (3.3) — FLT3-ITDhigh 115 (13.4) 6 (1.8) 71 (65.1) 38 (15.5) 0 — FLT3-ITD unknown 22 (2.6) 3 (1) 0 8 (3.7) 11 (6.1) — 16 MRC cytogenetic risk category MRC favorable 93 (10.8) 93 (28.4) 0 0 0 — MRC intermediate 551 (64) 215 (65.7) 98 (91) 80 (32.5) 158 (87.8) — MRC adverse 174 (20.2) 2 (0.6) 1 (1) 161 (65.4) 10 (5.5) — MRC not 43 (5) 17 (5.2) 9 (8) 5 (2.1) 12 (6.7) — classifiable ELN-2010 risk stratification3 Favorable 255 (29.6) 251 (76.8) 1 (1) 0 3 (1.7) — Intermediate I 260 (30.2) 38 (11.6) 78 (71.5) 44 (18) 100 (55.5) — Intermediate II 132 (15.3) 28 (8.6) 26 (23.8) 19 (7.8) 59 (32.8) — Adverse 178 (20.7) 1 (0.3) 0 177 (72.2) 0 — Not classifiable 36 (4.21) 9 (2.7) 4 (3.7) 5 (2) 18 (10) — BM, bone marrow; PB, peripheral blood. *P value of the comparison among ELN 2017 classifiable patients. induction chemotherapy (Figure 3), together with age and leukocyte 2017 stratification was maintained in patients above and below count at diagnosis. There was no significant difference in terms of 60 years (supplemental Material). induction mortality among the 3 groups. The multivariate analysis confirmed the independent prognostic value of the ELN risk classification (Figure 3). Thus, the hazard Relapse risk, NRM, EFS, and OS according to ELN ratio for death and event defined by EFS increased by ELN risk 2017 risk classification category, being 2.2- and 2.4-fold in intermediate-risk patients The median follow-up of the entire cohort (N 5 861) was 38 and 4.5- and 4.2-fold in adverse-risk patients. Interestingly, age, months. Detailed information regarding OS, EFS, and cumula- sex, and WBC at diagnosis also showed an independent prog- tive incidence of relapse (CIR) and NRM is summarized in nostic impact on OS and EFS, with a protective effect of female Table 2 and Figure 4. CIR and NRM at 5 years increased from sex. 28% and 7% to 40% and 14% and to 53% and 21% in favor- Outcome after alloSCT according to ELN able, intermediate, and adverse ELN categories, respectively (P , .001). This marked prognostic impact of the ELN 2017 2017 category category is translated into OS and EFS, with a gradual decline In the CETLAM-12 protocol, alloSCT policy varied according to the at 5 years from 70% and 60% to 46% and 31% and to 23% genetic risk category. AlloSCT was recommended in CR1 for all eligi- and 15%, for favorable-, intermediate-, and adverse-risk ble nonfavorable risk patients, after receiving at least 1 consolidation patients, respectively (P , .001). The prognostic impact of ELN course. In contrast, alloSCT was deferred in favorable-risk patients 1196 BATALLER et al 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 225 214 Favorable ELN 2017 risk Intermediate ELN 2017 risk 200 Adverse ELN 2017 risk No ELN 2017 class 180 175 150 Number of patients 125 100 96 75 71 52 50 47 30 25 22 21 18 17 18 14 13 11 14 7 10 5 1 gh s v 1) ) t ut PA 1) C ) 2) 5 3) ** 53 * * –7 ed in 16 as ;9 fa gh ** /– N ut m ;2 hi ;1 ;2 3; t(6 ut TP ng LN cl EB LN 6; hi q) m t(8 n. t(9 t(9 /t( X1 D m TD o (1 ra l(5 ab tE T E 1 C – 3) N N 53 ar –I )/t XL +/ –I de ut v( 1w ut RU ic re T3 16 TP 1m AS in e m et T3 PM yp A FL v( en lic M FL T2 ot in ut og N le P KM ry m t+ N al yt ka Bi 1 C 1w PM p. PM om N C N Figure 1. Distribution of patients with AML included in the study within the ELN 2017 risk categories. *Patients allocated to this category have been identified by NPM1 and FLT3-ITD mutational status. **Patients with ASXL1 mutation without RUNX1 mutation. ***Patients with TP53 mutation or abn(17p) without complex karyotype. after frontline treatment failure, either molecular or morphologic. Main Identification of a very-adverse-risk subgroup within characteristics of alloSCT are summarized in the supplemental Mate- the adverse ELN 2017 category rial. Of note, transplant characteristics were similar among the differ- ent ELN 2017 groups, with the exception of disease status at the Considering the wide biological heterogeneity within the ELN 2017 time of alloSCT. A higher proportion of patients from the adverse-risk adverse-risk category, we analyzed the outcome of the genetic enti- category received alloSCT without being in CR at the time of trans- ties included (supplemental Material). Indeed, some AML subtypes plant (14% vs 6%; P 5 .04). Concerning conditioning intensity and of the ELN 2017 adverse-risk category showed an exceptionally donor type, 48% of patients received a myeloablative regimen, and poor prognosis (very adverse risk, ELN Adv1), differentially worse an unrelated donor was used in 45% of transplant procedures. As a than that of the remaining patients of the adverse-risk category (ELN consequence of this risk-adapted transplant policy, median time from Adv2). These very-adverse-risk subtypes were as follows: AML with CR to alloSCT was similar for intermediate- and high-risk categories inv(3)(q21.3q26.2) or t(3;3)(q21.3q26.2), leading to deregulation of (2.9 and 3.1 months, respectively), and longer for favorable-risk GATA2 and MECOM (EVI1) genes (n 5 18), and AML with patients (11.7 months). mutated TP53 and/or 17p abnormalities and AML with complex kar- yotype (n 5 119, 49%; supplemental Figure 9.1). The complex kar- OS after alloSCT was similar in favorable- and intermediate-risk yotype subgroup included a high number of patients with del(17p)/ patients and worse in adverse-risk patients (OS at 2 years of 71% 217, del(5q)/25, and/or monosomal cytogenetic abnormalities and 67% in favorable- and intermediate-risk vs 47% in adverse-risk (supplemental Table 8; supplemental Figure 8.1). ELN Adv1 patients; P , .001). Relapse risk increased with ELN 2017 risk cat- patients had a higher proportion of male patients and presented egory, whereas NRM after transplant was comparable among with a lower leukocyte count and bone marrow blast infiltration at groups (Figure 5; Table 3). diagnosis. 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 VALIDATION OF THE AML ELN 2017 RISK STRATIFICATION 1197 Induction deaths (n=29) Death 8.9% (n=10) AML CR1 ELN 2017 Induction Maintained CR (n=295) Fav (n=327) (n=170) 90.2% (n=327) Mol relapse (n=40) Other (n=13) Refractory PR/Ref (n=4) AlloSCT Relapse CR (n=3) (n=82) (n=62) (n=55) 25% CR2 (n=25) CR2 (n=33) Relapse after NRM AlloSCT (n=14) (n=13) Induction deaths NRM (n=13) (n=10) 11.9% Death (n=1) AML ELN 2017 CR1 AlloSCT Induction Maintained CR Int (n=87) (n=80) (n=109) CR1(n=69) (n=51) (n=109) 79.8% 73.4% CR2 (n=7) Relapse after Refractory AlloSCT Relapse CR2 (n=9) (n=19) (n=13) (n=7) Ref (n=3) Ref (n=1) Induction deaths (n=25) NRM Death 10.2% (n=34) (n=5) AML CR1 AlloSCT ELN 2017 Induction Maintained CR (n=173) (n=150) Adv (n=245) (n=63) 70.6% CR1(n=125) 61.2% (n=245) CR2 (n=3) Relapse after Refractory AlloSCT Relapse CR2 (n=47) (n=53) (n=41) (n=5) Ref (n=16) Ref (n=6) Figure 2. Treatment disposition of patients according to responses and outcomes of patients of each ELN 2017 risk category. Patients included in the ELN Adv1 subgroup showed a lower pro- of patients amenable to alloSCT in CR1 (46% vs 75% for ELN portion of responders to frontline therapy (60% vs 81% for ELN Adv1 and ELN Adv2, respectively; P , .001), and a higher CIR Adv1 and ELN Adv2, respectively; P , .001), a lower proportion (75% vs 35% at 2 years for ELN Adv1 and ELN Adv2, 1198 BATALLER et al 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 Table 2. Outcomes at 2 and 5 years of the overall study population, and detailed for each ELN 2017 risk category Overall survival Event-free survival Cumulative incidence of Cumulative incidence of % (SE) % (SE) relapse % (SE) non-relapse mortality % (SE) All patients 2y 54.2 (1.8) 2y 41.7 (1.8) 2y 35.1 (1.9) 2y 10.8 (1.2) 5y 44.4 (2) 5y 35.9 (1.9) 5y 39.1 (2.1) 5y 13.6 (1.5) ELN 2017 favorable 2y 77.3 (2.4) 2y 66.5 (2.7) 2y 22.7 (2.6) 2y 3.5 (1.1) 5y 70.3 (3) 5y 59.6 (3.1) 5y 27.7 (3) 5y 6.8 (1.8) ELN 2017 intermediate 2y 51.8 (5.2) *** 2y 36.7 (5) *** 2y 36.9 (5.7) *** 2y 10.4 (3.5) *** 5y 45.7 (6.1) 5y 30.6 (5.9) 5y 39.5 (6.1) 5y 14.2 (5.1) ELN 2017 adverse 2y 32.5 (3.2) 2y 18 (2.7) 2y 51.5 (4.1) 2y 18.5 (3.1) 5y 23.1 (3.2) 5y 15.4 (2.7) 5y 52.6 (4.1) 5y 20.7 (3.4) ***P-value , .001; SE, Standard error. respectively; P , .001). As a consequence, OS and EFS were sig- and warrant the design of specific novel therapeutic approaches nificantly worse in the ELN Adv1 group, compared with Adv2, with aimed to improve the outcome of these patients. 2- and 5-year estimates of 15% and 10% vs 49% and 36% for OS Patients allocated to the ELN 2017 favorable category were con- (P , .001) and 3% (2- and 5-year) vs 33% and 28% for EFS (P , firmed to have a relatively good outcome, with a 5-year OS and CIR .001), respectively. This different outcome between subgroups was of 71% and 28%, respectively. Although alloSCT was not pre- confirmed regardless of the age range analyzed (,60 vs $60 planned in this subgroup, following previous recommendations,17,18 years) and was more evident in the younger patient population (2- approximately one quarter of patients received an alloSCT, mostly and 5-year OS: 61% and 50% vs 24% and 15% in ELN Adv2 vs because of molecular failure. This preemptive intervention based on ELN Adv1, respectively; P , .0001; supplemental Figure 9.5). MRD kinetics, might result in an overall clinical benefit. This had been Moreover, this different outcome was maintained in patients who suggested by a previous study performed by our group in ELN- could receive an alloSCT in CR1, with a 2-year OS after alloSCT of favorable patients with an NPM1 mutation, showing an improved out- 28% vs 58% and CIR at 2 years of 61% vs 21% for patients of come for patients treated at the time of molecular failure status com- the ELN Adv1 and ELN Adv2, respectively, without significant dif- pared with hematologic relapse.17 A similar observation has been ference in terms of NRM (Figure 6C-D). Additional comprehensive described in patients with core-binding factor–rearranged AML, in information comparing ELN Adv1 and ELN Adv2 is provided in the whom a preemptive intervention could lead to a survival improve- supplemental Material. Interestingly, when ELN Adv2 and ELN ment.18 Because most patients in this group harbor a molecular Adv1 categories were analyzed together with the remaining ELN marker amenable to quantitative MRD monitoring (ie, quantitative 2017 risk groups, the outcome of the ELN Adv2 group overlapped detection of mutated NPM1 or RUNX1/RUNXT1 and CBFb/MYH11 with that of ELN 2017 intermediate-risk group, whereas the out- fusion genes), a preemptive MRD-driven policy could be offered to come of ELN Adv1 was markedly poorer than the remaining cate- most patients in this category.19 This preemptive strategy translated gory. Thus, 4 prognostic groups could be clearly distinguished, into a favorable EFS of 67% and relatively low relapse incidence of although the ELN 2017 intermediate-risk group and the ELN Adv2 23% at 2 years, presumably because this preemptive intervention, group did not show statistically significant differences regarding OS which included an allo-SCT for molecular failure in 40 patients, pre- or EFS (Figure 6). vented the emergence of an overt hematologic relapse in a significant proportion of these patients. Interestingly, we also confirmed our pre- Discussion vious observation regarding the comparable outcome of patients with NPM1 mutation without FLT3-ITD or those with a FLT3-ITD mutation In this study, we validated the prognostic ability of the current AML with a low allelic ratio, supporting their inclusion in this favorable cate- ELN risk classification in a cohort of patients treated under the gory and reinforcing our MRD surveillance policy to prevent overt same risk-adapted protocol (CETLAM-12). This protocol prospec- relapses10 (supplemental Material). A recent study performed in the tively assigned a postremission strategy based on the genetic risk Cancer and Acute Leukemia Group B cooperative group (CALGB) evaluation at diagnosis, largely resembling that proposed by ELN has proposed a refined ELN classification according to specific 2017. Despite this risk-adapted post-CR strategy, ELN categories comutation patterns. Thus, the presence of WT1 mutations in segregated groups of patients with a markedly diverse outcome, NPM1-mutated AML or SETBP1 and BCOR in non-core binding with a 5-year OS of 70%, 46%, and 23% in favorable, intermediate factor rearranged (CBF) ELN-favorable patients was associated with and high-risk patients, respectively, mainly as a result of the different a worse outcome, and the authors proposed the reallocation of these initial response rate and relapse risk. Remarkably, we were able to patients to the adverse- and intermediate-risk group, respectively.20 identify a high-risk subset within ELN adverse category, the ELN Whether an MRD-driven preemptive treatment might overcome the very adverse risk (ELN Adv1), with a significantly poorer outcome, negative prognostic impact of these markers is unknown. with a 5-year OS of less than 10%. On the contrary, the outcome of the remaining ELN unfavorable patients not included in this very- As expected, ELN-defined adverse-risk features identified AML enti- high-risk subset showed a comparable outcome to that of ELN ties incurable for most patients, with a lower response rate to initial intermediate-risk patients. These results support a refinement of the chemotherapy, a 5-year OS of only 23%, and a relapse risk as high ELN classification with the recognition of this very-poor-risk subset as 53%. These patients represent a current clinical challenge with 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 VALIDATION OF THE AML ELN 2017 RISK STRATIFICATION 1199 Univariate and multivariate analysis for complete response (CR) Univariate Multivariate Variable OR (95% CI) p value OR (95% CI) p value ELN 2017 favorable Ref – Ref – ELN 2017 intermediate 0.43 (0.24–0.78) 0.005 0.5 (0.27–0.94) 0.03 ELN 2017 adverse 0.26 (0.17–0.41) 0.001 0.24 (0.14–0.38) 0.001 Age (10 year increase) 0.79 (0.67–0.92) 0.003 0.79 (0.66–0.94) 0.009 Sex (female) 1.31 (0.89–1.93) 0.172 1.34 (0.88–2.05) 0.167 WBC (25·109/L increase) 0.9 (0.84–0.97) 0.005 0.87 (0.80–0.94) 0.001 0.25 0.5 0.75 1 1.5 2 Lower odds for CR Higher odds for CR Univariate and multivariate analysis for overall survival (OS) Univariate Multivariate Variable HR (95% CI) p value HR (95% CI) p value ELN 2017 favorable Ref – Ref – ELN 2017 intermediate 2.37 (1.67–3.34) 0.001 2.19 (1.52–3.16) 0.001 ELN 2017 adverse 3.97 (3.05–5.18) 0.001 4.49 (3.41–5.92) 0.001 Age (10 year increase) 1.28 (1.16–1.40) 0.001 1.32 (1.19–1.46) 0.001 Sex (female) 0.71 (0.55–0.88) 0.003 0.64 (0.5–0.80) 0.001 WBC (25·109/L increase) 1.08 (1.04–1.12) 0.001 1.12 (1.08–1.16) 0.001 1 2 4 6 Higher risk of death Univariate and multivariate analysis for event free survival (EFS) Univariate Multivariate Variable HR (95% CI) p value HR (95% CI) p value ELN 2017 favorable Ref – Ref – ELN 2017 intermediate 2.55 (1.88–3.45) 0.001 2.36 (1.72–3.22) 0.001 ELN 2017 adverse 3.81 (3.01–4.81) 0.001 4.15 (3.26–5.29) 0.001 Age (10 year increase) 1.18 (1.09–1.28) 0.001 1.19 (1.09–1.29) 0.001 Sex (female) 0.83 (0.68–1.01) 0.064 0.78 (0.63–0.96) 0.021 WBC (25·109/L increase) 1.07 (1.03–1.11) 0.001 1.10 (1.06–1.14) 0.001 0.5 1 2 4 Higher risk of event Figure 3. Univariate and multivariate analysis for CR, OS, and EFS. OR, odds ratio; HR, hazard ratio; CI, confidence interval. the need to explore new drugs.21 The category is, however, highly genes. Accordingly, we identified a very-poor-risk subset, defined by heterogeneous, with diverse genetic lesions with distinct leukemo- the presence of inv(3)/t(3;3) with GATA2-MECOM1 rearrangement genic mechanisms, such as KMT2A rearrangements, a DEK- or a complex karyotype, highly enriched in cases with loss of chro- NUP98 fusion gene, or point mutations of TP53, ASXL1, or RUNX1 mosomal material at 5q, 7q, and/or 17p, presumptively involving 1200 BATALLER et al 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 A B 100% Overall survival 100% ELN 2017 favorable Event-free survival ELN 2017 intermediate 80% ELN 2017 adverse 80% Overall survival 60% Survival 60% 40% 40% 20% 20% p value < 0.0001 ELN fav vs ELN int p < 0.0001 ELN fav vs ELN adv p < 0.0001 0% 0% ELN int vs ELN adv p = 0.003 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Time (months) Time (months) Number at risk Number at risk 861 555 369 273 191 142 98 43 18 327 251 192 155 109 89 60 26 9 861 446 281 219 152 115 82 35 16 109 63 38 23 15 9 5 1 0 245 123 61 40 29 19 14 7 1 C D p value < 0.0001 Death w/o relapse ELN fav vs ELN int p < 0.0001 ELN 2017 favorable 50% Relapse ELN fav vs ELN adv p < 0.0001 ELN 2017 intermediate 100% ELN 2017 adverse ELN int vs ELN adv p = 0.015 40% 80% Cumulative incidence Event-free survival 30% 60% 20% 40% 10% 20% CI death p = 0.001 0% 0% CI relapse p = 0.001 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 Time (months) Time (months) Number at risk Number at risk 327 228 165 139 99 80 56 24 9 295 221 164 133 96 75 51 109 46 27 15 8 5 3 0 0 245 75 34 21 14 10 8 3 0 E F 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% CI death ELN 2017 fav vs ELN 2017 int p = ns CI death ELN 2017 fav vs ELN 2017 adv p = <0.001 CI death ELN 2017 int vs ELN 2017 adv p = ns CI relapse ELN 2017 fav vs ELN 2017 int p = 0.03 0% CI relapse ELN 2017 int vs ELN 2017 adv p = ns 0% CI relapse ELN 2017 fav vs ELN 2017 adv p = <0.001 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Time (months) Time (months) Number at risk Number at risk 87 51 30 18 11 6 4 173 70 38 26 17 12 8 Figure 4. Outcome (OS, EFS, CIR) of the entire cohort and in each ELN risk category: (A) OS and EFS of the entire cohort. (B) OS according to ELN 2017 risk category. (C) EFS according to ELN 2017 risk category. (D-F) CIR and death without relapse of patients allocated in the ELN 2017 favorable (D), intermediate (E), and adverse (F) risk category. Pairwise comparison of OS, EFS, and CI has been adjusted with the Bonferroni method. 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 VALIDATION OF THE AML ELN 2017 RISK STRATIFICATION 1201 A B 100% ELN 2017 favorable 100% ELN 2017 favorable ELN 2017 intermediate ELN 2017 intermediate 80% ELN 2017 adverse ELN 2017 adverse 80% Event free survival Overall survival 60% 60% 40% 40% p value < 0.001 p value < 0.001 20% ELN fav vs ELN int p = ns 20% ELN fav vs ELN int p = ns ELN fav vs ELN adv p = 0.003 ELN fav vs ELN adv p  0.001 ELN int vs ELN adv p = 0.04 ELN int vs ELN adv p = 0.03 0% 0% 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Time (months) Time (months) Number at risk Number at risk 82 54 34 29 19 15 7 1 1 82 51 33 28 18 15 7 1 1 80 44 23 18 11 6 4 1 0 80 43 22 16 10 5 4 1 0 150 74 47 36 23 15 7 3 1 150 63 38 31 21 14 6 3 1 C D E CIR post−alloSCT CIR post−alloSCT CIR post−alloSCT ELN favorable ELN intermediate ELN adverse Death w/o relapse 50% 50% 50% Relapse 40% 40% 40% Probability of an event CI death ELN 2017 fav vs ELN 2017 int p = ns CI death ELN 2017 int vs ELN 2017 adv p = ns CI death ELN 2017 fav vs ELN 2017 adv p = ns CI relapse ELN 2017 int vs ELN 2017 adv p = ns 30% 30% 30% CI relapse ELN 2017 fav vs ELN 2017 int p = ns CI relapse ELN 2017 fav vs ELN 2017 adv p < 0.001 20% 20% 20% 10% 10% 10% CI death p = ns CI relapse p = 0.003 0% 0% 0% 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Time (months) Time (months) Time (months) Number at risk Number at risk Number at risk 82 51 33 28 18 15 7 80 43 22 16 10 5 4 150 63 38 31 21 14 6 Figure 5. Outcome after alloSCT according to ELN 2017 risk category. (A) OS after alloSCT according to the ELN 2017 risk categories. (B) EFS after alloSCT according to the ELN 2017 risk categories. (C-E) CI of relapse and death without relapse of patients allocated in the ELN 2017 favorable (C), intermediate (D), and adverse (E) risk category. Pairwise comparison of OS, EFS, and CIR has been adjusted with the Bonferroni method. Starting time for the analysis is the date of alloSCT. TP53 mutation in many cases. This subgroup (ELN Adv1) showed fraction of patients from the ELN Adv1 category. In fact, both AML a 5-year OS of less than 10%. A high proportion of these patients, subgroups contained in this ELN Adv1 represent distinct biological more than 50%, do not achieve a planned alloSCT because of lack entities, and their particularly poor outcome has been previously rec- of response, and those who proceed to alloSCT have a high relapse ognized. A large part of the leukemogenic process driven by the risk after transplant. When we reanalyzed this adverse-risk category, GATA2-MECOM (EVI1) rearrangement, leading to GATA2 haploin- split between this very-adverse-risk subgroup (ELN Adv1) and the sufficiency and MECOM (EVI1) deregulated expression, has been remaining cases of the adverse-risk cohort, the outcome of both well characterized.22-25 This entity has been associated with a very adverse-risk groups was markedly different, and in fact, the ELN poor outcome, and the limited role of alloSCT in this context has Adv2 group overlapped with the patients with intermediate-risk been specifically analyzed.22,26,27 On the other hand, the poor prog- AML (Figure 6). This observation suggests that the treatment plan- nosis of complex karyotype in AML seems, in part, explained by the ning according to our protocol, with an alloSCT early after CR, could frequent coexistence of TP53 multihit mutations inducing a loss of benefit patients from the ELN Adv2 subgroup but only a small the functional wild-type TP53 protein, leading to chemoresistance, 1202 BATALLER et al 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 Table 3. Outcomes after alloSCT according to the ELN 2017 risk stratification Cumulative incidence of death without relapse after OS after alloSCT, % (SE) EFS after alloSCT, % (SE) CIR after alloSCT, % (SE) alloSCT, % (SE) All patients 2 y, 58.2 (2.6) 2y, 52.5 (2.7) 2y, 26.7 (2.3) 2y, 20.8 (2.1) ELN 2017 favorable 2 y, 70.5 (5.4)* 2 y, 66.6 (5.6)* 2 y, 17 (4.6)† 2 y, 16.4 (4.2)NS ELN 2017 intermediate 2 y, 67.4 (5.7) 2 y, 58.5 (6.4) 2 y, 29 (6.1) 2 y, 12.5 (4) ELN 2017 adverse 2 y, 46.9 (4.5) 2 y, 40.8 (4.4) 2 y, 36.1 (4.2) 2 y, 23.1 (3.6) NS, not significant; SE, standard error. *P , .001. †P , .01. as recently recognized in myelodysplastic syndromes.28 Unfortu- groups.7,20 Because the use of FLT3 inhibitors was minimal in this nately, as a limitation of our study, we were not able to determine patient cohort, the impact of the generalized use of midostaurin (or the proportion of patients from the complex karyotype/TP53 subset alternative FLT3 inhibitors) should be reanalyzed in future studies. that corresponded to a biallelic TP53 mutation. Similarly, the adverse These targeted FLT3 inhibitors have been shown to improve out- outcome even after alloSCT of frequent cytogenetics such as 25/ comes in patients with FLT3-mutated AML and could lead to the del(5q), 27/del(7q), 17p, and/or TP53, in the context of a complex need of risk group reallocation in the future.39 In addition, this cate- karyotype, has been directly addressed and confirmed in recent gory includes patients with cytogenetic abnormalities not classified studies.29-31 Given the poor outcome obtained with the standard as favorable or adverse, as well as patients without NPM1 mutation approach based on intensive AML-type chemotherapy and alloSCT, and adverse-risk genetic lesions, which are biologically heteroge- the recognition of this subset with very poor prognosis could be neous.29,40 In this context, further knowledge on the prognostic essential for the design of novel therapeutic approaches. Potential impact of additional markers and gene mutation interactions, such innovative strategies to improve outcomes of this AML subgroup as the deleterious effect associated with BCOR, SETBP1, ZRSR2, should focus on pretransplant treatment to increase the response or DNMT3A comutations might contribute to redefine the current rate and prevent early relapse, as well as maintenance after alloSCT classification in future updates, as suggested by Eisfeld et al20 from to enhance the graft-versus-leukemia effect and avoid posttransplant the CALGB network.41 relapse.32 Novel current available agents, approved for AML, could The ELN 2017 classification was proposed by a panel of experts contribute to improve outcome of high-risk patients obtained with refining the previous ELN 2010 prognostic stratification, collapsing intensive chemotherapy. Specifically, 1:5 daunorubicin/cytarabine the 2 intermediate subgroups into a unique category and including liposomal formulation CPX-351 has demonstrated its role in elderly the contemporary evidence on individual markers with independent patients diagnosed with AML with myelodysplasia-related changes, prognostic value in different studies.3,4 Its validation has been retro- with a clear impact on those patients bridged to alloSCT after treat- spectively performed in a few studies that confirmed, with some limi- ment with this agent.33 Similarly, the combination of venetoclax with tations, the ELN 2017 clinical value.7-9,20 These studies differ in the azacytidine has been established as the standard of care in patients patient population included (up to 60 years in 3 of these 4 studies) not candidate to intensive chemotherapy; its role in younger patients, and the treatment protocol used, and, most importantly, they did not combined with intensive chemotherapy and other agents, is now include a ELN category–based transplant policy. Nonetheless, all been explored.34,35 More recently, investigational agents such as these studies were able to confirm its prognostic impact and pro- eprenatopopt (APR-246) magrolimab (monoclonal antibody directed posed some refined definitions. Herold et al7 also identified a very against macrophage immune checkpoint CD47), or flotetuzumab poor outcome among patients with TP53 mutation and complex kar- (dual-affinity re-targeting [DART], antibody molecule to CD3 and yotype and highlighted the negative impact of DNMT3A mutations CD123) have shown promising activity in patients with very high-risk within every ELN 2017 risk group. The aforementioned study by Eis- genetic features, including TP53 mutation, and should be explored feld et al20 reallocated specific subgroups of patients according to in therapeutic algorithms for higher-risk patients.36-38 specific comutations (eg, BCOR and SETBP1 in non-CBF favor- The current intermediate-risk group represents the smallest cate- able AML and IDH-mutated adverse risk patients should be rede- gory, involving less than 20% of all patients, being highly enriched fined as intermediate-risk AML; WT1 and ZRSR comutation in in FLT3-ITD. This low proportion of intermediate-risk patients can be NPM1-mutated AML confer an adverse prognosis). With our analy- attributed, in part, to the nonclassifiable group composed of patients sis, we contribute to validate the ELN prognostic stratification in a not harboring a cytogenetic or PCR-based genetic ELN defining prespecified treatment protocol, for patients up to 70 years eligible category and without available NGS testing. Presumably, and based to intensive chemotherapy with a clearly predefined alloSCT policy. on other large series providing accurate information on the fre- In conclusion, the current ELN 2017 risk classification is able to quency of RUNX1, ASXL, and TP53 sole mutations, the largest part segregate groups of patients with a different prognosis and is useful of this nonclassifiable subgroup must correspond to intermediate- to guide postremission alloSCT indications. Moreover, the current risk patients. With the assumption of 70% of nonclassifiable analysis confirms that these outcome differences among ELN cate- patients as intermediate-risk patients, approximately 29% of our gories persist even after predefined, risk adapted postremission study population would correspond to the ELN intermediate-risk cat- strategies, establishing alloSCT in first complete remission for all egory, a proportion similar to that found in other large cooperative suitable patients with a nonfavorable risk. The poor outcome 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 VALIDATION OF THE AML ELN 2017 RISK STRATIFICATION 1203 A B 100% Favorable risk 100% Favorable risk Intermediate risk Intermediate risk Adverse risk Adverse risk 80% Very adverse risk (Adv+) 80% Very adverse risk (Adv+) Event-free survival Overall survival 60% 60% 40% 40% 20% 20% p value < 0.001 p value < 0.001 Fav vs Int, Fav vs Adv and Fav vs Adv+ p  0.001 Fav vs Int, Fav vs Adv and Fav vs Adv+ p  0.001 Int vs Adv p = ns and Int vs Adv+ p  0.001 Int vs Adv p = ns, and Int vs Adv+ p  0.001 0% Adv vs Adv+ p  0.001 0% Adv vs Adv+ p  0.001 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 80 Time (months) Time (months) Number at risk Number at risk 327 251 192 155 109 89 60 26 327 228 165 139 99 80 56 24 9 109 63 38 23 15 9 5 1 109 46 27 15 8 5 3 0 0 126 77 45 33 24 17 13 6 126 59 29 20 14 10 8 3 0 119 46 16 7 5 2 1 1 119 16 5 1 0 0 0 0 0 C D 100% Adv– 80% Adv+ 70% 80% 60% 50% Overall survival 60% 40% 40% 30% 20% 20% 10% 0% 0% 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 Time (months) Time (months) Number at risk Relapse ELN Adv– NRM ELN Adv– 95 54 38 29 20 13 6 2 1 Relapse ELN Adv+ NRM ELN Adv+ 55 20 9 7 3 2 1 1 0 Figure 6. Outcome according to ELN risk category dividing the adverse risk category in two different subgroups, ELN Adv2 and the very adverse risk/ELN Adv1. (A) OS and (B) EFS of the ELN 2017 risk groups stratifying the ELN 2017 adverse risk patients in the 2 proposed groups (ELN Adv2 and very adverse risk/ELN Adv1). Pairwise comparison of OS and EFS has been adjusted with the Bonferroni method. Outcome after alloSCT in both ELN Adv subgroups in terms of OS (C) and CIR and NRM (D). observed in adverse-risk patients, and especially among ELN Adv1, novel therapeutic approaches that might overcome the limitations of despite alloSCT, warrants the investigation of experimental strate- the current standard plan. gies aimed to improve pretransplant response depth and prevent relapse after transplantation. Additionally, an accurate clinical use of Acknowledgments MRD monitoring can complement the information provided by This study was supported (in part) by Fondo de Investigaciones genetic characterization at diagnosis and identify patients from the Sanitarias/Instituto de Salud Carlos III (ISCIII) grants PI16/01027, favorable-risk category who can benefit from an alloSCT in CR1. PI19/1476, and PI20/01621; Health Research and Innovation Finally, our study confirmed the existence of a very-poor-risk subset Strategic Plan (PERIS) grant SLT002/16/00433 and research among adverse-risk patients who deserve specific investigation of group support SGR 1395 and SGR 1655 from Generalitat de 1204 BATALLER et al 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 Catalunya; resident award “Emili Letang” 2019 (Hospital Clınic de consultancy for Abbvie, Novartis, Astellas, Pfizer, and Jazz Barcelona); and “Beca de Investigacion FEHH 2019” (Fundacio n Pharmaceuticals and grand support from Novartis and Jazz Espa~ nola de Hematologia y Hemoterapia). Pharmaceuticals. All other authors declare no competing finan- cial interests. Authorship ORCID profiles: A.B., 0000-0002-6085-2745; F.G., 0000- Contribution: A.B., A.G., F.G., G.O., M.D.-B., M.A., M.T., S.V., 0001-6376-0929; G.O., 0000-0003-2180-2371; R.C., 0000- M.P.Q.d.L., R.C., D.G., F.V.-L., L.E., A.G.-G., O.S., A.S., B.M.M., J.B., 0003-0560-1254; F.-V.L., 0000-0002-0718-5955; A.S., 0000- S.C.-D, and A.O.-C. collected the clinical data. M.L.-G., M.P., L.Z., 0001-7465-6203; B.M.M., 0000-0002-6733-6006; S.C.-D., 0000- J.F.N., D.C., M.R., and D.C. performed the cytogenetic and molecu- 0001-9044-4665; D.E., 0000-0001-8897-3170; A.O-C., 0000- lar tests. A.B., G.O., M.D.-B., F.G., A.R., P.M., S.B., J.S., and J.E. 0002-7921-5420; A.R., 0000-0002-4233-5031; P.M., 0000- analyzed the data and wrote the manuscript. All authors reviewed 0001-9528-4971; M.P., 0000-0001-6375-596X; J.E., 0000-0002- and accepted the manuscript. 8056-648X. Conflict-of-interest disclosure: J.S. reports advisory and Correspondence: Alex Bataller, Hematology Department, Hospi- speaker fees for Abbvie, Astellas, Jazz Pharmaceuticals, Novar- tal Clınic of Barcelona, Barcelona, Spain; e-mail: abataller@clinic. tis, and CSL Behring and grant support from Astellas, Novartis, cat; or Jordi Esteve, Hematology Department, Hospital Clınic of Amgen, Jansen, Pfizer, and Jazz Pharmaceuticals. J.E. reports Barcelona, Barcelona, 08036 Spain; e-mail: jesteve@clinic.cat. References 1. D€ ohner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med. 2015;373(12):1136-1152. 2. Cheson BD, Bennett JM, Kopecky KJ, et al; International Working Group for Diagnosis, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia. Revised recommendations of the International Working Group for diagnosis, standardization of response criteria, treatment outcomes, and reporting standards for therapeutic trials in acute myeloid leukemia. J Clin Oncol. 2003;21(24):4642-4649. 3. D€ohner H, Estey EH, Amadori S, et al; European LeukemiaNet. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet. Blood. 2010;115(3):453-474. 4. D€ ohner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. 5. R€ ollig C, Bornh€ auser M, Thiede C, et al. Long-term prognosis of acute myeloid leukemia according to the new genetic risk classification of the European LeukemiaNet recommendations: evaluation of the proposed reporting system. J Clin Oncol. 2011;29(20):2758-2765. 6. Mr ozek K, Marcucci G, Nicolet D, et al. Prognostic significance of the European LeukemiaNet standardized system for reporting cytogenetic and molecular alterations in adults with acute myeloid leukemia. J Clin Oncol. 2012;30(36):4515-4523. 7. Herold T, Rothenberg-Thurley M, Grunwald VV, et al. Validation and refinement of the revised 2017 European LeukemiaNet genetic risk stratification of acute myeloid leukemia. Leukemia. 2020;34(12):3161-3172. 8. Boddu PC, Kadia TM, Garcia-Manero G, et al. Validation of the 2017 European LeukemiaNet classification for acute myeloid leukemia with NPM1 and FLT3-internal tandem duplication genotypes. Cancer. 2019;125(7):1091-1100. 9. Harada Y, Nagata Y, Kihara R, et al; Japan Adult Leukemia Study Group JALSG. Prognostic analysis according to the 2017 ELN risk stratification by genetics in adult acute myeloid leukemia patients treated in the Japan Adult Leukemia Study Group (JALSG) AML201 study. Leuk Res. 2018;66(66):20-27. 10. Pratcorona M, Brunet S, Nomded eu J, et al; Grupo Cooperativo Para el Estudio y Tratamiento de las Leucemias Agudas Mieloblasticas. Favorable outcome of patients with acute myeloid leukemia harboring a low-allelic burden FLT3-ITD mutation and concomitant NPM1 mutation: relevance to post-remission therapy. Blood. 2013;121(14):2734-2738. 11. Gorello P, Cazzaniga G, Alberti F, et al. Quantitative assessment of minimal residual disease in acute myeloid leukemia carrying nucleophosmin (NPM1) gene mutations. Leukemia. 2006;20(6):1103-1108. 12. Benthaus T, Schneider F, Mellert G, et al. Rapid and sensitive screening for CEBPA mutations in acute myeloid leukaemia. Br J Haematol. 2008; 143(2):230-239. 13. Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials. 1996;17(4):343-346. 14. Delgado J, Pereira A, Villamor N, L opez-Guillermo A, Rozman C. Survival analysis in hematologic malignancies: recommendations for clinicians. Haematologica. 2014;99(9):1410-1420. 15. Medeiros BC. Interpretation of clinical endpoints in trials of acute myeloid leukemia. Leuk Res. 2018;68(January):32-39. 16. Grimwade D, Hills RK, Moorman AV, et al. Refinement of cytogenetic classification in AML Younger adult patients treated in UKMRC. Blood. 2010; 116(3):354-366. 17. Bataller A, O~ nate G, Diaz-Beya M, et al; Grupo Cooperativo Para el Estudio y Tratamiento de las Leucemias Agudas y Mielodisplasias (CETLAM). Acute myeloid leukemia with NPM1 mutation and favorable European LeukemiaNet category: outcome after preemptive intervention based on measurable residual disease. Br J Haematol. 2020;191(1):52-61. 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4 VALIDATION OF THE AML ELN 2017 RISK STRATIFICATION 1205 18. Yalniz FF, Patel KP, Bashir Q, et al. Significance of minimal residual disease monitoring by real-time quantitative polymerase chain reaction in core binding factor acute myeloid leukemia for transplantation outcomes. Cancer. 2020;126(10):2183-2192. 19. Schuurhuis GJ, Heuser M, Freeman S, et al. Minimal/measurable residual disease in AML: a consensus document from the European LeukemiaNet MRD Working Party. Blood. 2018;131(12):1275-1291. 20. Eisfeld A-K, Kohlschmidt J, Mims A, et al. Additional gene mutations may refine the 2017 European LeukemiaNet classification in adult patients with de novo acute myeloid leukemia aged ,60 years. Leukemia. 2020;34(12):3215-3227. 21. DiNardo CD, Wei AH. How I treat acute myeloid leukemia in the era of new drugs. Blood. 2020;135(2):85-96. 22. Lugthart S, Gr€oschel S, Beverloo HB, et al. Clinical, molecular, and prognostic significance of WHO type inv(3)(q21q26.2)/t(3;3)(q21;q26.2) and various other 3q abnormalities in acute myeloid leukemia. J Clin Oncol. 2010;28(24):3890-3898. 23. Gr€oschel S, Sanders MA, Hoogenboezem R, et al. A single oncogenic enhancer rearrangement causes concomitant EVI1 and GATA2 deregulation in leukemia. Cell. 2014;157(2):369-381. 24. Katayama S, Suzuki M, Yamaoka A, et al. GATA2 haploinsufficiency accelerates EVI1-driven leukemogenesis. Blood. 2017;130(7):908-919. 25. Yamaoka A, Suzuki M, Katayama S, Orihara D, Engel JD, Yamamoto M. EVI1 and GATA2 misexpression induced by inv(3)(q21q26) contribute to megakaryocyte-lineage skewing and leukemogenesis. Blood Adv. 2020;4(8):1722-1736. 26. Halaburda K, Labopin M, Houhou M, et al. AlloHSCT for inv(3)(q21;q26)/t(3;3)(q21;q26) AML: a report from the acute leukemia working party of the European society for blood and marrow transplantation. Bone Marrow Transplant. 2018;53(6):683-691. 27. Sitges M, Boluda B, Garrido A, et al; PETHEMA, CETLAM Cooperative Groups. Acute myeloid leukemia with inv(3)(q21.3q26.2)/ t(3;3)(q21.3;q26.2): Study of 61 patients treated with intensive protocols. Eur J Haematol. 2020;105(2):138-147. 28. Bernard E, Nannya Y, Hasserjian RP, et al. Implications of TP53 allelic state for genome stability, clinical presentation and outcomes in myelodysplastic syndromes [corrections published in Nat Med. 2021;27:562 and 927]. Nat Med. 2020;26(10):1549-1556. 29. Papaemmanuil E, D€ ohner H, Campbell PJ. Genomic classification in acute myeloid leukemia. N Engl J Med. 2016;375(9):900-901. 30. Mr ozek K, Eisfeld AK, Kohlschmidt J, et al. Complex karyotype in de novo acute myeloid leukemia: typical and atypical subtypes differ molecularly and clinically. Leukemia. 2019;33(7):1620-1634. 31. Britt A, Mohyuddin GR, McClune B, et al. Acute myeloid leukemia or myelodysplastic syndrome with chromosome 17 abnormalities and long-term outcomes with or without hematopoietic stem cell transplantation. Leuk Res. 2020;95:106402. 32. Thekkudan SF, Lima M, Metheny L. Prevention of relapse after allogeneic stem cell transplantation in acute myeloid leukemia: updates and challenges. Adv Cell Gene Ther. 2020;3(2):1-18. 33. Lancet JE, Uy GL, Cortes JE, et al. CPX-351 (cytarabine and daunorubicin) liposome for injection versus conventional cytarabine plus daunorubicin in older patients with newly diagnosed secondary acute myeloid leukemia. J Clin Oncol. 2018;36(26):2684-2692. 34. DiNardo CD, Jonas BA, Pullarkat V, et al. Azacitidine and venetoclax in previously untreated acute myeloid leukemia. N Engl J Med. 2020;383(7): 617-629. 35. DiNardo CD, Lachowiez CA, Takahashi K, et al. Venetoclax combined with FLAG-IDA induction and consolidation in newly diagnosed and relapsed or refractory acute myeloid leukemia. J Clin Oncol. 2021;39(25):2768-2778. 36. Cluzeau T, Sebert M, Rahm e R, et al. Eprenetapopt plus azacitidine in TP53-mutated myelodysplastic syndromes and acute myeloid leukemia: a phase II study by the Groupe Francophone des My elodysplasies (GFM). J Clin Oncol. 2021;39(14):1575-1583. 37. Uy GL, Aldoss I, Foster MC, et al. Flotetuzumab as salvage immunotherapy for refractory acute myeloid leukemia. Blood. 2021;137(6):751-762. 38. Sallman DA, Asch AS, Al Malki MM, et al. The first-in-class anti-CD47 antibody magrolimab (5F9) in combination with azacitidine is effective in MDS and AML patients: ongoing phase 1b results. Blood. 2019;134(suppl 1):569. 39. Stone RM, Mandrekar SJ, Sanford BL, et al. Midostaurin plus chemotherapy for acute myeloid leukemia with a FLT3 mutation. N Engl J Med. 2017; 377(5):454-464. 40. Eisfeld AK, Mr ozek K, Kohlschmidt J, et al. The mutational oncoprint of recurrent cytogenetic abnormalities in adult patients with de novo acute myeloid leukemia. Leukemia. 2017;31(10):2211-2218. 41. O~nate G, Bataller A, Garrido A, et al. Prognostic impact of DNMT3A mutation in acute myeloid leukemia with mutated NPM1 [published online ahead of print 13 September 2021]. Blood Adv. doi: 10.1182/bloodadvances.2020004136. 1206 BATALLER et al 22 FEBRUARY 2022 • VOLUME 6, NUMBER 4