Dear Editor,

Acute myeloid leukemia (AML) is a genetically heterogenous malignancy of the hematopoietic stem and progenitor cells that occurs primarily in older adults, with a median age of 69 years at diagnosis and around 60% of patients aged ≥ 60 years [1]. AML develops from the acquisition of somatic genomic alterations resulting in a clonal proliferation of undifferentiated immature precursors preventing normal hematopoiesis. Novel technologies including next-generation sequencing (NGS) have identified recurrent structural chromosomal abnormalities and (driver) gene mutations [2]. More than 95% of AML patients harbor at least one identifiable somatic mutation, some of which carry prognostic impact [3,4,5].

Lower-intensity therapy with hypomethylating agents and the addition of venetoclax or ivosidenib in patients with IDH1 mutations is the standard of care for newly diagnosed (ND) older adults and/or those who are unfit for intensive chemotherapy. Particularly in the older patient population, therapies targeting genetic mutations are attractive options as they are relatively well tolerated with fewer side effects. While the genomic landscape of AML in older patients has been shown to be different from that of younger patients, most studies were relatively small or did not include patients treated with lower-intensity treatment [5]. We performed an updated and more comprehensive analysis of cytogenetic and mutational data from over 1000 older patients with ND AML from a previously published smaller cohort of patients treated on the Beat AML clinical trial (NCT03013998) [6], aimed to characterize the incidence and co-occurence of genomic abnormalities present in AML patients aged ≥ 60 years.

This multicenter, retrospective study included patients aged ≥ 60 years with ND AML enrolled in the Beat AML clinical trial (NCT03013998) with a consent date before May 10, 2023. Details of the genomic analysis have been previously described [6]. Cytogenetic analyses from diagnostic samples were centrally reviewed. FLT3-ITD ratio was assessed using LeukoStrat CDx FLT3 Mutation Assay (Invivoscribe) and NGS was performed using FoundationOne®Heme (Foundation Medicine) with a limit of detection of 28 reads. Foundation Medicine uses the COSMIC database to categorize mutations as known or likely pathogenic or variants of unknown significance (VUS). VUS were included in the final analysis because their clinical implications are disparate. The presence of multiple hotspot mutations within one gene in an individual patient was counted as one gene mutation. Mutations with any detectable variant allele frequency (VAF) were considered as positive for a mutation. See supplement for full description of methods.

A total of 1023 patients aged ≥ 60 years diagnosed with AML were included in the analysis. The median age at time of diagnosis was 72 years (range, 60-92 years). Baseline characteristics are described in Table 1. Seventy percent of the patients were treated with lower-intensity therapy regimens. A summary of the treatment regimens is listed in Supplemental Table S1. Cytogenetic analysis was available for 926 (91%) patients. Karyotype was normal in 333 patients (36% of patients with known karyotypes) and complex in 280 patients (30%). Six percent had a -5/del(5q), -7/del(7q) or -17/abn(17p) abnormality without complex karyotype, and 2% had a KMT2A-rearrangement. Nearly half of the patients (44%) had myelodysplasia-related cytogenetic abnormalities, including complex karyotype, -5/del(5q)/add(5q), -7/del(7q), trisomy 8, del(12p)/t(12p)/add(12p), i(17q), -17/add(17p)/del(17p), del(20q) or idic(X)(q13).

Table 1 Baseline characteristics of the study patients (N = 1023).
Full size table

Mutation analysis was available for all patients. Overall, the ten most frequently mutated genes were IDH1/2 (n = 282; 28%), DNMT3A (n = 259; 25.3%), TP53 (n = 256; 25.0%), TET2 (n = 241; 23.6%), RUNX1 (n = 227; 22.2%), SRSF2 (n = 222; 21.7%), ASXL1 (n = 215; 21.0%), FLT3 (n = 210, 21.0%; FLT3-ITD (n = 123; 12.0%), FLT3-TKD (n = 79; 7.7%)), NPM1 (n = 207; 20.2%), and NRAS (n = 175; 17.1%) (Fig. 1A). Eight patients harbored a mutation in the CEBPA bZIP domain (0.8%). Myelodysplasia related-gene mutations, including ASXL1, BCOR, EZH2, RUNX1, SF3B1, SRSF2, STAG2, U2AF1, and ZRSR2 were detected in 57% of the patients.

Fig. 1: Gene mutation profile in patients 60 years and older with AML.
figure 1

A Common driver gene mutations present in ≥ 10% of the patients. Gene mutations with any detectable VAF were considered present. B Analysis of variant allele frequencies of the top 10 most frequently mutated genes. The boxplots show the median, 25th and 75th percentiles, minimum, maximum, and outliers. The black dashed line marks an allele frequency of 0.5. C Co-occurrence of molecular markers across molecular subgroups. The number at the top of each column is the number of patients belonged to the subclass. Red and blue highlighted boxes indicate significant (FDR-corrected p < 0.05) co-occurrence (red) or mutual exclusively (blue) between two gene mutations or between gene mutations and recurrent cytogenetic abnormalities. P-values were calculated using the Fisher’s exact test and were corrected for the number of patients with available cytogenetics (n = 926/1023). For example, of the 280 patients with complex karyotype, 213 patients also had a TP53 mutation, while 5 patients had a NPM1 mutation.

Full size image

Based on prior studies showing importance of these genes in the development and progression of AML, we compiled a list of 102 “driver” gene mutations in AML that were assessed in our study [5, 7]. Driver mutations had a median VAF of 0.44, lower than the VAF of heterozygous germline polymorphisms. Figure 1B illustrates the observed VAFs present in the 10 most frequently mutated genes in our cohort. FLT3 and NRAS had the lowest VAF, suggesting that alterations in these signaling pathways are often acquired relatively late during the evolution of the leukemic clone. The VAF of TP53 varied largely between patients (range, 0.01-0.99).

Integrated mutation analysis recognized frequently co-occurring or mutually exclusive mutations (i.e., both genes mutated in more/fewer patients than expected by their individual frequencies). A list of the most important significant co-occurrences and mutually exclusive combinations is shown in Fig. 1C and listed in Supplemental Table S2. Notable significant positive associations were identified between NPM1 with each of DNTM3A and FLT3-ITD (p < 0.001). FLT3-ITD alterations also often co-occurred with DNMT3A (p < 0.001). IDH2-R140 co-occurred with NPM1, and IDH2-R172 frequently co-occurred with DNMT3A but was mutually exclusive with NPM1. TP53 mutation did not significantly co-occur with any other gene mutation but was mutually exclusive or present below the expected frequency with several of the most commonly mutated genes. Similarly, a systematic analysis of all pairwise associations between cytogenetics and genes was performed. Core-binding factor (CBF) AML was mutually exclusive with ASXL1 (1% of the CBF patients had an ASXL1 mutation vs 22% overall, p = 0.048), DNMT3A (0% vs 24% overall, p = 0.003), RUNX1 (0% vs 22% overall, p = 0.001), TET2 (0% vs 23% overall, p < 0.001), and TP53 (0% vs 26% overall, p = 0.002), while they frequently co-occurred with KIT (34% vs 4% overall, p < 0.001) and NRAS (46% vs 17% overall, p < 0.001). DNMT3A (36% vs 14%), FLT3-ITD (21% vs 3%), NPM1(43% vs 2%), IDH1/2 (41% vs 11%), and TET2 (30% vs 15%) were more frequently mutated in normal karyotype then complex karyotype AML, respectively. In contrast, TP53 was mutated in 76% of patients with complex karyotype compared to 4% of those with normal karyotype (p < 0.001). A list of the most important significant co-occurrences and mutually exclusive combinations is shown in Fig. 1C and listed in Supplemental Table S3.

In this study we performed a comprehensive analysis of the presence, patterns of co-occurrence of gene mutations in a uniformly sequenced cohort of >1000 ND AML patients aged ≥ 60 years. This study is particularly relevant as only few studies have solely focused on older adults with AML and/or treated with lower-intensity treatment regimens [8,9,10,11], which contrasts the real world where the majority of patients with AML are 60 years or older [1].

Cytogenetic abnormalities were identified in 64% of the older AML patients, with a high incidence of adverse cytogenetic risk features, such as complex karyotypes, and chromosome 5, 7, and 17 abnormalities. About half of the AML patients had myelodysplasia-related cytogenetic abnormalities, which is higher than the previously reported [12]. Furthermore, we identified 15 driver gene mutations that were present in ≥ 10% of the patients, including in IDH1/2, DNMT3A, TP53, TET2, RUNX1, SRSF2, ASXL1, FLT3, NPM1, and NRAS, of which many with a distinct prevalence compared to studies evaluating younger patient cohorts (Supplemental Table S4). The detection of mutated TP53 in 25% of patients aged ≥ 60 years has only been shown in one other recently published study [8], with most studies reporting rates ranging from 2-9%. Moreover, mutations in myelodysplasia-related genes (e.g., ASXL1, BCOR, RUNX1, SF3B1, SRSF2, STAG2, U2AF2, and ZRSR2) that now confer a diagnosis myelodysplastic syndrome (MDS)/AML rather than MDS for patients with ≥ 10% blasts per Internation Consensus Classification [13], were detected in 57% of the patients in our cohort, which is more than twice the frequency in younger patients [14]. This highlights yet another adverse-risk disease marker in older AML patients.

The lower incidence of NPM1 and FLT3-ITD mutations in older adults compared to younger adults is consistent with previous findings, showing that 1) NPM1 peaks at middle age and decreases in patients > 70 years of age, and 2) that activating FLT3 mutations are less common in older patients with AML [15]. Finally, the incidence of mutations in DNMT3A, NRAS, PTPN11, IDH1, CEBPA, KIT, and KRAS was relatively similar across adult age groups. Mutations in signaling and kinase pathways lead to aberrant activation and proliferation of cellular signaling and are frequently identified in sub-clonal cellular fractions, indicating they are often late clonal events in disease evolution [5, 15].

To conclude, we demonstrated that AML in older patients is a molecularly heterogeneous disease with a genomic profile that is distinct from AML in patients younger than < 60 years. The high prevalence of TP53 and myelodysplasia-related gene mutations and cytogenetic abnormalities, likely impacts outcomes, irrespective of patient’s performance status and medical fitness. However, 48% of AML patients ≥ 60 years present with a mutation that is potentially targetable for treatment (i.e., FLT3, IDH1, IDH2, NPM1, KMT2A-rearrangement), underscoring the potential of targeted therapy options tailored to the genomic alterations present in this age group.