Acute Myeloid Leukemia in the Elderly: Molecular Abnormalities and Molecular Classification

Abstract

Acute myeloid leukemia (AML), a heterogeneous and aggressive clonal disease, is predominantly observed in older individuals, with a median age at diagnosis of 68–69 years. With the aging population, there is a significant increase in the occurrence of some genetic alterations, including detrimental gene mutations and cytogenetic abnormalities, and a higher incidence of secondary AML (s-AML) and therapy-related AML (t-AML), compared to younger AML patients. Outcomes of AML patients and their response to therapy are associated with the molecular features of AML subtypes and with individual variables. The current criteria for risk classification predict outcomes in younger AML patients treated with intensive chemotherapy but are less predictive for older AML patients treated with lower-intensity treatments. Thus, this review analyzes and discusses the development of new risk stratification models adapted to the study of older AML patients and how these new criteria may significantly contribute to a more rational classification and treatment of older AML patients.

1. Introduction

Acute myeloid leukemia (AML) is a malignant blood disorder characterized by the uncontrolled proliferation and blocked differentiation of hematopoietic stem/progenitor cells (HSPCs). The incidence of AML increases with age, with a peak occurring at a median age of 68 years and slightly decreasing in people aged over 75 years [1]. Aging is clearly associated with the leukemogenic process and a poor prognosis [2].

Three main types of AML can be distinguished at the clinical level, as follows: de novo AML, which is not related to a prior malignant or pre-malignant condition; secondary AML (s-AML), which emerges from prior hematologic malignancies, such as myelodysplastic syndromes (MDS) or myeloproliferative disorders, as well as from clonal hematopoiesis (CH), a premalignant condition commonly observed in aged individuals; and therapy-related AML (t-AML), resulting from previous exposure to chemotherapeutic agents or radiation.

Age is one of the main determinants of AML outcome, with 5-year survival for patients above 60 years not exceeding 10–15% [3]. This finding is supported by the observation that age is associated with the accumulation of adverse cytogenetic abnormalities [4,5].

Somatic mutations progressively accumulate in HPSCs during the natural aging process. It has been estimated that—by the age of 70—humans harbor between 350,000 and 1,400,000 coding mutations within the pool of HSPCs [6]. These mutations can provide a competitive growth advantage to cells and determine the development of clonal hematopoiesis in a healthy aging population [7]. DNMT3A, TET2, and ASXL1 genes are the most frequently mutated genes in CH; with advancing age, the relative proportion of DNMT3A-CH decreases, while TET2 and ASXL1 frequency increases. SRFSF2 mutations are rare before the age of 50 but clearly increase with advancing age [8].

Recent studies have shown some remarkable differences in the genetic landscape of older AML compared to younger AML, involving a higher incidence of TET2, DNMT3A, TP53, ASXL1, and SRSF2 mutations and a higher incidence of adverse cytogenetic abnormalities (5q/5 and 7 chromosome monosomies and complex karyotype) compared to younger AML [9,10].

A pangenomic analysis of a large cohort of older AML patients enabled the reconstruction of the sequence of mutation acquisition and the main oncogenic trees. This analysis showed the existence of five branches, with ASXL1, DDX41, DNMT3A, TET2, and TP53 as master genes emanating from the root [11]. The tree emanating from the ASXL1 node is complex and articulated, generating EZH2, NRAS, RUNX1, and UZAF1 as individual clones; the branches originating from the ASXL1 node include eight of the nine genes defined as myelodysplasia-related gene mutations [11]. The branches originating from the DNMT3A node first involve the acquisition of an NPM1 mutation and then FLT3-ITD or PTPN11 gene alterations [11]. The TET2 node generates only one branch containing CEBPA; the TP53 and DDX41 nodes do not generate further branching [11]. Hierarchical clustering allowed the classification of older AMLs into six different clusters. Cluster 1 (49%) represents the largest group and includes AML with myelodysplasia-related cytogenetic abnormalities without TP53 mutations; this group includes the largest proportion of NRAS and KRAS mutations. Cluster 2 (25%) includes complex karyotypes associated with TP53 mutations. Cluster 3 (17%) includes frequent NPM1, DNMT3A, and FLT3 mutations. Cluster 4 is characterized by structural and numerical chromosomal alterations. Cluster 5 is rare and is characterized by IDH2^R172 mutations. Cluster 0 (6%) includes three subgroups characterized by DDX41 alterations, MECOM (EVI1), and core binding factor leukemias [11].

Li and coworkers performed a comprehensive molecular analysis of 1474 newly diagnosed AML patients of different ages and explored some leukemogenetic processes associated with aging. During aging, gene fusions decrease, while coding mutations increase in the older age group of AML [12]. Based on the presence of clonal hematopoiesis-related mutations (DNMT3A, TET2, and ASXL1) and myelodysplasia-related mutations, these mutant subgroups were identified among gene fusion-negative AMLs: CH-AML (clonal hematopoiesis-AML, DNMT3A and TET2 mutations without secondary type mutations), CH-MDS-AML (clonal hematopoiesis-myelodysplasia, secondary-type mutations-positive with or without previous MDS history), and other gene fusion-negative AMLs (enriched in NPM1, RUNX1, and CEBPA mutations) [12]. Age and the presence of genetic mutations in TP53 or ETV6 predict a poor prognosis in CH-AML; male gender, WBC count, mutations in RUNX1, IKZF1, TP53, FLT3, and spliceosome genes, as well as monosomal karyotype, are associated with a poor outcome in CH-MDS-AML patients [12].

A recent study reported the molecular characterization of 1063 older AML patients (median age 72 years), 9 of whom were enrolled as part of the Beat AML study [13]. The most frequent mutations were IDH1-2 (28%), DNMT3A (25%), TP53 (25%), TET2 (23%), RUNX1 (22%), SRFSF2 (22%), ASXL1 (21%), FLT3-ITD (12%), FLT3-TKD (7.7%), nPM1 (20%), and NRAS (17%) [13]. MDS-related mutations were observed in 57% of patients [13]. Importantly, analysis of the VAFs of recurrent mutant genes showed values greater than 0.4 for genes such as DNMT3A, TET2, ASXL1, and SRSF2, around 0.4 for RUNX1, NPM1, and IDH1-2, and lower values for FLT3 and NRAS; finally, VAF values for the TP53 mutant gene ranged from <0.3 to >0.7 [13].

These three molecular classifications of AMLs have made an important contribution because they have provided support for the analysis of older AMLs, contributing to the identification of molecularly relevant subgroups of patients suitable for targeted therapy.

The 2022 ELN classification assigns an adverse risk to MDS-R mutations without any favorable genetic changes, and this group represents about 30% of total AML [14]. Myelodysplasia-related (MDS-R) mutations were observed in 44.9% of older patients compared to 23.4% of younger patients; in both older and younger AML patients, MDS mutations had a negative prognostic impact on response to treatment with intensive chemotherapy [15]. Meclenbrauck et al. explored a group of 550 AMLs with MDS-R mutations of different ages [16]. A higher proportion of MDS–mutated older AML patients pertain to the ELN 2022 adverse risk and have a high variant allele frequency of one or more mutated MDS-related genes, a higher WBC at diagnosis, and a significantly higher frequency of ASXL1 mutations [16]. In the adverse risk group, a high VAF of MDS-R mutations was associated with lower survival than in the MDS-R group with low VAF, after intensive chemotherapy and also after allo-HSCT [16].

Shimony et al. explored a group of 314 older AML patients (median age 74 years); 111 patients had TP53 mutations, 115 had secondary AML, as supported by the presence of MDS-associated SRSF2, U2AF1, SF3B1, and ASXL1 mutations, and 89 had de novo AML [17]. These patients were treated with HMA alone or in combination with VEN, with mOS values of 7.4 and 9.9 months, respectively, in the whole AML population. No benefit was observed among TP53-mutant patients when comparing HMA + VEN to HMA treatment (5.7 vs. 6.1 months, respectively), or among de novo AML patients (13.2 vs. 10.3 months, respectively). A clear benefit was observed in s-AML patients (14.1 months vs. 6.9 months, respectively) [17]. These findings clearly show that in older AML patients treated with HMA + VEN, the prognostic impact of many genetic alterations is different from that observed in younger AML patients treated with intensive chemotherapy [17]. Furthermore, the analysis of a large cohort of 395 s-AML patients showed that the presence of KRAS/NRAS co-mutations was associated with a decreased response, and the presence of SF3B1 mutations was associated with an increased response to HMA + VEN [18].

2. Risk Stratification of Older AML Patients Not Suitable for Intensive Chemotherapy

The current risk stratification systems available for AML patients were based on younger AML patients undergoing intensive chemotherapy treatment. Thus, it is not surprising that these systems fail to accurately predict the risk in elderly AML patients undergoing treatment with VEN and HMA.

An open-label, non-randomized phase Ib trial (NCT02203773) and the confirmatory VIALE-A phase III randomized trial (NCT02993523), both enrolling patients with intermediate and poor-risk cytogenetics, according to NCCN guidelines and not suitable for intensive chemotherapy treatment, showed that the combination of VEN plus azacitidine (AZA) induced greater overall survival compared to azacitidine alone (mean OS 14.7 months vs. 9.6 months, respectively). The analysis of these patients, subdivided into three risk groups, showed that the median OS was similar for the favorable- and intermediate-risk groups who received VEN + AZA (21.09 months vs. 23.26 months, respectively); the mOS for the high-risk group was lower (11.53 months), but this group was composed of two subgroups with the following distinct outcomes: one with TP53 mutations (mOS of 5.42 months) and the other with RUNX1 mutations (22.9 months) [19]. These patients were also analyzed using the ELN 2022 criteria, showing that the mOS with VEN + AZA versus placebo + Aza was 39 months vs. 11 months, respectively, in the favorable-risk group; 15.2 vs. 9.1 months in the intermediate-risk group, and 12.7 months vs. 9.3 months in the adverse-risk group [19]. However, in the VEN + AZA-treated patients, OS outcomes were similar for patients with intermediate-risk and high-risk AML (mOS values of 15.2 months and 12.7 months, respectively) [19].

Given these limitations of the ELN 2017 and ELN 2022 classification systems to predict the outcomes of AML patients treated with VEN + AZA, a prognostic tool was developed based on the analysis of 29 genetic alterations, including seven genes not considered in ELN 2022, such as TET2, DNMT3A, IDH1, IDH2, FLT3-TKD, KRAS and NRAS [17]. The patients treated with VEN + AZA could be categorized into one of the three VEN + AZA benefit groups based on the analysis of the mutational status of four of these genes: TP53, FLT3-ITD, NRAS, and KRAS (molecular prognostic risk signature, mPRS). The group of patients with the highest benefit included those with TP53-WT, NRAS, KRAS-WT, and no FLT3-ITD mutations. The intermediate-benefit group had TP53-WT as well as NRAS/KRAS and FLT3-ITD mutations; the lower-benefit group had TP53 mutations; the mOS values of these three groups were 26.5, 12.1, and 5.5 months, respectively [20]. Figure 1 clearly shows that the mPRS classification allowed for better stratification of mOS in the three groups identified as favorable, intermediate, and adverse risk.

The observations made in this study were further confirmed through the retrospective analysis of 159 AML patients treated with VEN and an HMA [21]. Using the mPRS model, these patients were stratified into high-benefit, intermediate-benefit, and low-benefit groups, exhibiting a significantly different mOS and mEFS [21]. The molecular analysis of these three groups allowed us to define some remarkable differences. For example, patients in the lower-benefit group had a lower incidence of NPM1, ASXL1, RUNX1, and IDH1/2 mutations than patients in the high- and intermediate-benefit groups; therapy-related AMLs were more frequent among low-benefit patients; secondary AMLs were more frequent among high- and intermediate-benefit groups than in the low-benefit group (Figure 2). In conclusion, this study showed that TP53 (low benefit) and NRAS/KRAS (intermediate benefit) negatively affect outcomes of AML patients treated with Ven + HMA, as shown in Figure 2.

Other recent studies have confirmed that TP53-mutant AMLs are scarcely responsive to VEN + HMA or HMA alone, while secondary AMLs display a significant improvement in OS when treated with VEN + HMA compared to HMA alone [17]. This observation is in line with the study by Bataller et al. [21], which showed that secondary mutations (SRFSF2, ASXL1) are significantly more frequent in the high- and intermediate-benefit groups compared to the low-benefit group.

Other studies have shown the limited prognostic predictive capacity of the ELN 2017 and ELN 2022 risk classifications in older AML patients treated with less-intensive regimens [22]. In fact, a study carried out in 604 older AML patients treated as part of a phase III clinical study found that both ELN 2017 and ELN 2022 classified most of these patients in the adverse-risk prognostic group (62% to 73%), with only a minority of patients included in the favorable-risk group (13–17%); both risk classifications failed to provide significant separation of the survival curves, with the favorable- and intermediate-risk curves largely overlapping [11]. To improve the risk stratification of these patients, Jahn and coworkers explored the clinical parameters and genetic abnormalities with a clear impact on survival, showing a clear impact of WBC and ECOG status and DDX41, TP53, and FLT3-ITD gene abnormalities. [11]. AML with mutated DDX41 has a clearly favorable OS, while AML with TP53 or FLT3-ITD mutations has a similarly poor outcome, and AML without these genetic abnormalities has an intermediate outcome [11].

Given the limitations of ELN 2017 and ELN 2022 in predicting the risk of older AML patients undergoing non-intensive chemotherapy treatments, a group of experts recently proposed a new genetic risk classification for these patients [22]. This classification was based on the analysis of clinical studies carried out using HMA monotherapy, HMA plus VEN, and HMA plus IDH inhibitors. Thus, an ELN genetic risk classification was proposed for patients with newly diagnosed AML undergoing less-intensive treatment based on HMA [20]. According to this classification, the mOS for patients with favorable risk was estimated to be >24 months; for those with adverse risk, between 5 and 8 months; and for those with intermediate risk, 12–18 months [22]. According to this classification, the adverse-risk group includes TP53-mutated AML; the intermediate-risk group includes FLT3-ITD–positive and/or NRAS-mutated and/or KRAS-mutated TP53-WT cases; the favorable group includes five different AML subtypes, as follows: NPM1-mutated (no-FLT3-ITD, NRAS-WT, KRAS-WT and TP53-WT), IDH2-mutated (no-FLT3-ITD, NRAS-WT, KRAS-WT and TP53-WT), IDH1-mutated (TP53-WT), DDX41-mutated, and other molecular abnormalities (no-FLT3-ITD, NRAS-WT, KRAS-WT and TP53-WT) [22]. This risk classification provides an important background for the consensus stratification of older patients receiving less-intensive treatments. However, the ELN 2024 less-intensive recommendations need to be validated through clinical trials and real-world studies. A recent real-world study in UK patients treated with VEN + AZA (587 patients) or with low-dose cytarabine (67 patients) showed that the new ELN 2024 classification performed better than ELN 2022 [23].

Finally, for more comprehensive risk assessment, it is recommended to integrate the ELN 2024 evaluations with the assessment of minimal residual disease [22]. In fact, recent studies have shown the strong prognostic value of MRD assessment in NPM1-mutated AML patients receiving VEN-based treatment [24].

The Beat AML program evaluated a cohort of 595 older AML patients treated with low-intensity treatments and proposed a refinement of the ELN 2022 classification [22]. In the ELN 2022 classification, the large majority of these patients were classified as adverse-risk (78%), and only a minority were classified as favorable-risk (11%) and intermediate-risk (11%); however, ELN 2022 failed to stratify favorable-risk from intermediate-risk AML patients [25]. Multivariate analysis showed that IDH2 mutations act as an independent favorable prognostic factor, while KRAS, MLL2, and TP53 mutations act as unfavorable prognostic factors [25]. For each of these mutations, a mutation score was developed and evaluated for each combination of these mutations (−1 for IDH2 mutations and +1 for each of the KRAS, MLL2, and TP53 mutations); a score of ≤0 points was defined as “Beat AML intermediate” and ≥1 point was defined as “Beat AML adverse” [25]. Introducing these criteria, the ELN risk stratification was refined, as follows: The Beat ELN favorable group was composed of ELN 2022 favorable + ELN 2022 intermediate (21.3% of total); the Beat ELN intermediate group was composed of ELN 2022 adverse-risk patients with a score of 0 (40% of the total) and the Beat ELN adverse group was composed of ELN 2022 adverse-risk patients with a score of ≥1 point [25]. Since the whole population of patients was treated with different low-intensity treatments, a subanalysis performed on patients treated with VEN + HMA confirmed the capacity of Beat AML 2024 to discriminate among three groups of patients, namely, favorable-, intermediate-, and adverse-risk [25]. The mutational profile showed a higher frequency of TP53, MLL2, and KRAS mutations and chromosome 5 and 7 abnormalities in Beat AML adverse-risk AMLs, while NPM1 and IDH2 mutations were more frequent in Beat AML favorable-risk AMLs, as shown in Figure 3 [25].

A recent study—through the reclassification of some mutations and incorporation of new gene mutations in the ELN2024 risk stratification model—improved survival prognostication following VEN-based therapy [26]. The evaluation of newly diagnosed older AML patients treated with VEN plus an HMA provided additional criteria for an improved risk classification, defining three groups of AML patients, as follows: A favorable-risk group (mutated NPM1, IDH1, IDH2, DDX41, and wild-type NRAS, KRAS, PTPN11, FLT3-ITD, and TP53), an intermediate-risk group (mutated FLT3-ITD, NRAS, and other mutations not classified, and wild-type KRAS, PTPN11, and TP53), and an adverse-risk group (mutated KRAS, PTPN11, or TP53) [23] (Figure 4). This revised ELN2024 prognostic evaluation of older AML patients undergoing treatment with VEN and an HMA allowed better risk stratification of favorable- and intermediate-risk AML groups [26] (Figure 4).

In a recent study, Hoff et al. retrospectively analyzed a multicenter cohort of 238 patients enrolled in the Beat AML clinical trial with newly diagnosed AML aged 60 years or older, treated with HMS + VEN, with the aim of comparing the 4-gene ELN2024, the refined ELN2024, and the Beat AML risk stratification models [27]. Although the Beat AML risk stratification better discriminates the favorable, intermediate, and adverse risk groups than the ELN 2024 stratification models, none of these models was able to predict outcomes in all patients [27].

Table 1. Comparative analysis of the AML subdivision into favorable, intermediate, and adverse risk groups, following ELN 2022, ELN 2024, and refined ELN 2024.

	ELN 2022	ELN 2024	Refined ELN 2024
Favorable	NPM1-mut (without FLT3-ITD) Cytogenetic abn [t(8;21), Inv(16), t(15;17)]	IDH2-mut IDH1-mut NPM1-mut DDX41-mut FLT-3-WT, KRAS-WT, NRAS-WT, TP53-WT	IDH2-mut IDH1-mut NPM1-mut DDX41-mut FLT-3-WT, KRAS-WT, NRAS-WT, TP53-WT, PTPN11-WT
Intermediate	FLT3-ITD Cytogenetic abn and molecular abn not classified as favorable or adverse	FLT3-ITD KRAS-mut NRAS-mut TP53-mut Other cytogenetic and molecular abn	FLT3-ITD NRAS-mut KRAS-WT PTPN11-WT TP53-WT Other cytogenetic and molecular abn
Adverse	Some cytogenetic abn Complex karyotype -5, -7, -17 Mutated ASXL1, BCOR, EZH2, RUNX1, SF3B1, SRSF2, STAG2, U2AF1, ZRSR2 TP53-mut	TP53-mut	TP53-mut KRAS-mut PTPN11-mut

presents a comparative analysis of the ELN 2022, ELN 2024, and refined ELN 2024 AML classifications. The most notable differences between ELN 2024 and ELN 2022 involve the following: (i) the classification of IDH1 and IDH2 mutations, which are included in the favorable-risk group in ELN 2024, while they are classified in the various risk groups in the ELN 2022; the classification of MDS-R mutations, which are classified in the adverse-risk group in ELN 2022, while they are classified in different risk groups in ELN 2024 (Table 1). The most relevant differences between ELN 2024 and refined ELN 2024 concern KRAS and PTPN11 mutations, which are included in the adverse risk group in refined ELN 2024, while KRAS mutations are included in the intermediate risk group, and PTPN11 mutations are not considered in ELN 2024 (Table 1).

Du et al. explored 209 elderly/unfit AML patients undergoing treatment with venetoclax and hypomethylating agents, and developed a clinical prediction model based on ten clinical, biochemical, and genetic parameters [28]. Factors associated with drug sensitivity include the hyperdiploid or polyploid karyotype, and mutations in IDH, NPM1, and CEBPA; factors associated with drug resistance include FAB classification (M4/M5; M0/M6/M7), s-AML, t-AML, R/R -AML, adverse-risk karyotype, mutations in ASXL1 and FLT3 [28]. This model focused on predicting drug sensitivity more than drug resistance, incorporating a broader range of predictive variables, including NPM1, CEBPA, and IDH mutations [28].

Other studies have contributed to defining genetic variables that define resistance to VEN + HMA therapy. Gangat et al. explored 103 older AML patients (median age 74 years) undergoing VEN + HMA treatment; TP53 and FLT3-ITD mutations were enriched among patients not responding to this treatment, while ASXL1 mutations were associated with an increased response rate [29]. However, ASXL1 mutations were associated with a high level of MRD positivity and a high rate of relapse [29]. Age-adjusted survival analysis showed that NPM1 and IDH1/2 mutations and CR are favorable-risk factors, while TP53 and ASXL1 mutations are unfavorable-risk factors [29]. According to the presence or absence of adverse risk factors, a simple risk point model was generated [29]. A second, more extensive study by the same authors, based on the retrospective analysis of 301 older AML patients undergoing VEN and HMA combination therapy, provided the identification of the factors correlated with a favorable or unfavorable response [30]. A total of 60% of these patients achieved a CR, and 69% were classified in the adverse group according to the ELN risk classification [30]. Multivariate analysis identified NPM1, IDH2, and DDX41 mutations as favorable predictors of CR, and TP53, FLT3-ITD, and RUNX1 mutations as unfavorable predictors [30]. CR rates varied from 36% in the presence of unfavorable mutations and absence of favorable mutations, to 91% in the presence of favorable mutations and absence of unfavorable mutations [30]. Risk factors for reduced survival included failure to achieve CR, adverse-risk karyotype, TP53 mutation, and the absence of IDH2 mutations; these risk factors were used to develop an HR-weighted risk model that predicts outcomes better than the ELN genetic risk model [30].

Another study developed a prognostication risk model for AML patients undergoing treatment with VEN + HMA; this study was based on the retrospective analysis of 212 older AML patients treated with VEN + HMA [31]. The aim of this study was to identify factors predicting primary resistance to VEN + HMA in elderly AML patients [31]. Using a multivariate logistic regression model, factors associated with VEN + HMA resistance were FAB M5 (AML with monocytic differentiation), MDS s-AML, RUNX1, and FLT3-ITD mutations [31]. Using these findings, an eight-point scoring system that allowed the stratification of AML patients into three groups was developed; non-remission rates of low-, intermediate-, and high-risk groups were 22%, 60% and 77%, respectively [31].

A recent study further supported the monocytic differentiation features of AMLs associated with resistance to VEN + HMA therapy, through a mechanism related to the preferential expression, in these leukemias, of anti-apoptotic proteins such as Mcl1 and BCL2A1, and to the presence of mutations such as KRAS mutations [32].

3. Genetic Risk Classification of Older AML Patients Suitable for Treatment with Intensive Chemotherapy

Some older AML patients may be treated with intensive induction chemotherapy and receive some benefit from this treatment. Thus, it is fundamental to have clinical and biological criteria for the identification of older AML patients suitable for 7 + 3 induction chemotherapy.

Few studies have explored the genetic and clinical levels of older AML patients who are candidates for intensive chemotherapy treatment.

Itzykson et al. assessed the predictors of short-term (remission) and long-term (overall survival) benefits from intensive chemotherapy in a large group of patients aged over 60 years with AML, based on the ALFA 1200 cytogenetic and sequencing study. According to ELN 2017, 41%, 27%, and 28% of these patients pertained to the adverse, intermediate, and favorable-risk groups, respectively; according to cytogenetic profiling, the patients were subdivided into good (2.8%), intermediate (72%), and poor (17.8%) groups, with a clearly lower OS in the poor cytogenetic group compared to the good/intermediate groups [33]. The mutational profile showed that the most recurrently mutated genes were DNMT3A (28.7%), NPM1 (27.0%), TET2 (21%), and FLT3-ITD (18.7%, with an allelic ratio of ≥0.5 in 24% of FLT3-ITD-mutated cases) [33]. In univariate analysis, NPM1 mutations were associated with a lower hazard ratio for death, while mutations in DNMT3A, NRAS, ASXL1, RUNX1, PHF6, CSF3R, SETBP1, and ETV6 conferred a higher risk of death; only FLT3-ITD with an allelic ratio ≥0.5 worsened prognosis; TP53 and KRAS mutations were associated with a higher hazard ratio for death [33]. In multivariate analysis, in patients with non-poor cytogenetics, NPM1 mutations predicted improved OS, while FLT3-ITD, DNMT3A, NRAS, and ASXL1 mutations independently predicted a shorter OS; in patients with poor cytogenetics, TP53 and KRAS mutations independently predicted a worse OS [29]. According to the HR observed, a scoring system was generated, assigning −1 point for an NPM1 mutation, +1 point for FLT3-ITD with a low allelic ratio, DNMT3A, NRAS, and ASXL1, and 2 points for FLT3-ITD with a high allelic ratio [33]. Patients with non-poor cytogenetics and either an NPM1 mutation with at most 1 mutation among mutations in FLT3-ITD (low allelic ratio), DNMT3A, ASXL1, or NRAS, or those with NPM1, FLT3-ITD, DNMT3A, ASXL1, or NRAS, all wild-type, were assigned to a favorable prognostic group (“go-go” group); patients with adverse-risk cytogenetics and either a mutation in KRAS or TP53 were assigned to an adverse-risk group (“no-go” group); the rest of the patients were assigned to an intermediate-risk group “slow-go” group) [33]. Thus, this decision tool enabled good discrimination among the three groups with significantly different OS profiles [33].

Versluis et al. recently reported on the development of a novel risk stratification system for patients 60 years and older (AML60+) who are eligible for intensive induction chemotherapy [34]. The ELN 2022 classified these leukemias into favorable-risk (4yr OS 53%), intermediate (4yr OS 31%), and adverse-risk (4yr OS 18%). An extensive evaluation of these patients identified nine variables at diagnosis predicting survival, each evaluated using a scoring system correlated with the individual HR of each of these variables, as follows: mutated TP53 and monosomal karyotype were assigned three points; age >65 years, two points; WBC > 20 × 10⁹/L, male sex, FLT3-ITD, DNMT3A, ASXL1, and RUNX1 mutations, each received one point [30]. Using the AML60+ scoring system, four groups were defined as favorable (0–1 points, 14%), intermediate (2–3 points, 40%), poor (4–6 points, 28%) and very poor (7–10 points, 17%); these four groups of patients corresponded to patients with progressively decreasing OS, decreasing CR rate, and decreasing rate of transplantation [21]. Importantly, each of the ELN 2022 risk groups is heterogeneous with respect to the AML60+ risk groups—the ELN 2022 favorable-risk group contains 75% of intermediate- or higher-risk patients, and the ELN 2022 adverse-risk group contains 37% of AML60+ favorable- and intermediate-risk patients [34]. Allo-HSCT was associated with improved survival in the intermediate and very poor groups, but not in the favorable and poor groups; in fact, in these last two groups, the benefits deriving from non-relapse mortality are offset by increased non-relapse mortality induced by allo-HSCT [34].

4. Outlines of the Current Strategies for the Treatment of Older AML Patients

The treatment of older AML patients remains particularly challenging. A significant proportion of these patients are unfit for intensive therapy, and just a few years ago, the treatment of these patients was mainly limited to supportive care. However, the introduction of hypomethylating agents, such as AZA and decitabine, enabled the development of new therapeutic opportunities for these patients. In particular, the development of the HMA + VEN regimen has become an important therapeutic option for older AML patients unfit for intensive chemotherapy.

Upon hospital admission, it is essential that newly diagnosed older AML patients are carefully evaluated for risk stratification through standard clinical and laboratory criteria, morpho-cytogenetic analysis, and next-generation sequencing to assess the mutational profile. Assigning intensive or non-intensive treatment based on the fitness of older AML patients represents the first therapeutic choice for these patients; this decision-making process may be challenging and is generally driven by patient- and disease-related characteristics.

It is important to note that, according to the Beat AML Master study, it was estimated that about 48% of older AML patients bear molecular abnormalities that can be targeted using specific pharmacologic agents; these AMLs include patients with IDH1, iDH2, FLT3, and NPM1 mutations, or with KMT2A rearrangements [35].

Patients who are fit for intensive chemotherapy undergo treatment with intensive induction chemotherapy alone or associated with targeted therapy in molecularly suitable patients; patients with MDS-R gene alterations may alternatively receive treatment with HMA + VEN [36]. Patients who achieve complete remission and are eligible for transplantation may receive allo-HSCT, and those who are transplant-ineligible receive consolidation chemotherapy and/or alternative therapies (Figure 5).

Patients unfit for intensive chemotherapy may receive hMA + VEN as the best low-intensity treatment, eventually in combination with targeted therapy in molecularly eligible patients. Recent analyses of clinical trials and real-world studies support the efficacy of HMA + VEN, comparable to that observed for fit older AML patients treated with intensive chemotherapy and superior to that observed with other low-intensity treatments. Responding patients in the ELN-2024 favorable-risk group may continue therapy with or without HSCT; responding patients in the ELN-2024 intermediate- and adverse-risk groups may undergo allo-HSCT if transplant-eligible, while transplant-ineligible patients may continue therapy until progression [36].

Treating older patients with adverse-risk AML is very challenging. These leukemias are mainly represented by TP53-mutant AML. For TP53-mutant AML, standard therapies, such as intensive induction chemotherapy, as well as HMA + VEN, all lead to a similar and dismal mOS of 6–9 months [37]. Due to its inherent resistance to therapy, international guidelines recommend a clinical trial as the first approach to the treatment of TP53-mutant AML. In the absence of investigational trials, it is recommended that these patients be treated with HMA + VEN if they are fit for allo-HSCT; intensive chemotherapy if they have TP53 mutations with low VAF and monoallelic disease; or HMA or HMA + VEN if they are unfit for allo-HSCT [33].

5. Conclusions

The treatment of older AML patients remains particularly challenging, and their overall survival remains poor. This is due not only to the reduced fitness of the majority of these patients, but also to the severity of the disease.

Recent studies have shown some peculiarities of AMLs observed in older patients compared to younger adult patients. The current risk stratification models of AML are largely derived from studies performed in younger patients, and their application to older patients is unsatisfactory. However, recent studies based on the analysis of large cohorts of elderly AML patients treated with VEN + HMA have led to the definition of new risk stratification models suitable for more accurate prognostication of these patients.

It is fundamental and clinically relevant to identify, for each individual older AML patient, according to clinical and biological (genetic alterations) criteria, whether intensive chemotherapy, targeted therapy, or VEN/HMA is the most suitable option. These choices require the development of efficient and validated risk-stratification criteria for these patients and a better understanding of the molecular abnormalities observed in older AML patients. These new risk-stratification criteria may allow us to identify some “super-responder” patients within the higher-benefit subset who may be able to discontinue therapy with VEN + HMA after a finite duration. Furthermore, it is important to identify patients who have an elevated probability of being refractory to VEN + HMA therapy and who have very poor outcomes. Finally, the inclusion of MRD evaluation represents a valuable tool to further improve the predictive value of these new stratification systems.