To the Editor

According to the international consensus classification (ICC) nine mutations define acute myeloid leukemia (AML) with myelodysplasia-related gene (MRG) mutations: ASXL1, BCOR, EZH2, RUNX1, SF3B1, SRSF2, STAG2, U2AF1 and ZRSR2 [1]. Patients with MRG mutations have been shown to be older, leukopenic at diagnosis and overall have a worse outcome [2]. The 2022 European LeukemiaNet (ELN) classification assigns an adverse risk to patients with MRG mutations without any favorable genetic changes [3], who constitute up to 30% of AML patients [4,5,6,7,8].

Several studies have debated the prognostic impact of MRG mutations both with [9,10,11] and without favorable co-mutations [5,6,7, 12, 13] rendering heterogeneous results. As MRG mutations are acquired at an early stage of disease evolution and further mutations are acquired over the course of the disease [2], we hypothesized that considering clonality could provide a better understanding of the prognostic impact of MRG mutations.

Patients ≥ 18 years with newly diagnosed AML were included, who were diagnosed between 2000 and 2021 and underwent intensive induction treatment followed by either allogeneic hematopoietic cell transplantation (alloHCT) or consolidation chemotherapy. Patients from 15 academic centers in Germany were considered if molecular data from a myeloid panel, analyzed at Hannover Medical School, cytogenetic, and clinical data were available. At the time of diagnosis blood or bone marrow samples were analyzed via next generation sequencing using a 46 or 48 gene panel (Supplementary Table S1) as described before [14]. ASXL1 c.1934 dupG was excluded if the variant allele frequency (VAF) was <18% based on internal validation. For mutations located on the X-chromosome the VAF was adjusted for hemizygosity in men.

Of the 550 cases identified, 153 patients (28%) were classified as ELN favorable risk, 158 (29%) as intermediate risk, and 239 (43%) as adverse risk (Supplementary Fig. S1).

Among ELN favorable patients, 31 (20%) had at least one MRG mutation. The mutational landscape is described in Supplementary Tables S2, S3.

Patients with MRG mutations within the ELN favorable risk group had similar baseline characteristics as favorable risk patients without MRG mutations (Supplementary Table S4). After a median follow-up of 5.4 years, patients in the favorable risk group had similar event-free (EFS) (HR 1.04, 95%CI 0.59–1.83, p = 0.88) and overall survival (OS) (HR 1.26, 95%CI 0.68–2.34, p = 0.46) with or without MRG mutations (Supplementary Fig. S2A, B). This also held true in multivariate analysis considering baseline risk factors and all mutations found in at least 10 patients (Supplementary Fig. S3).

A total of 255 MRG mutations were detected in 156 of the 239 patients in the ELN adverse risk group (Supplementary Tables S2, S3). Patients with MRG mutations showed a significantly worse EFS (HR 1.25, 95%CI 1.08–1.46, p = 0.003) and by trend OS (HR 1.19, 95%CI 1.0–1.42, p = 0.054) compared to ELN favorable risk patients (Supplementary Fig. S4A, B) but a similar EFS and OS compared to ELN intermediate risk patients (EFS: HR 1.19, 95%CI 0.9–1.59, p = 0.2; OS: HR 1.19, 95%CI 0.85–1.67, p = 0.3, Supplementary Fig. S4A, B). MRG mutated patients also displayed a similar EFS compared to the ELN adverse risk group (HR 0.84, 95%CI 0.61–1.16, p = 0.3), while OS was significantly longer compared to the ELN adverse risk group without MRG mutations (HR 0.61, 95%CI 0.43–0.87, p = 0.007; Supplementary Fig. S4A, B).

Next, maximally selected rank statistics were used to determine the most prognostic threshold of the VAF in MRG mutations, which was found at 44.5% (Fig. 1A) [15]. This VAF cut-off was similar to the median VAF of all MRG mutations (41.9%). For simplicity we implemented a cut-off at a VAF of 45% for further analysis. First, the prognostic effect of a high or low VAF of MRG mutations was explored in the ELN favorable risk group. The 16 patients (10%) with low VAF and the 15 patients (10%) with high VAF had similar OS and EFS compared to the 122 ELN favorable risk patients without MRG mutation (Supplementary Fig. S5A, B).

Fig. 1: Prognostic impact of the size of the MRG mutated clone.
figure 1

A Among the 156 patients bearing at least one MRG mutation and classified as ELN adverse risk, the highest VAF was chosen to calculate a prognostically relevant VAF cut-off using maximally selected rank statistics. The most discriminating cut-off was 45%. B Event-free and C overall survival of patients with at least one MRG mutation with a VAF ≥ 45% compared to patients where all MRG mutations have a VAF < 45%.

Full size image

Among the patients classified as ELN adverse risk, 91 (38%) patients in whom all MRG mutations had a VAF below 45% were classified as low VAF patients, while 65 (27%) patients with at least one MRG mutation with a VAF ≥45% were classified as high VAF patients. High VAF patients were significantly older and had a higher white blood cell (WBC) count and lower platelet counts at diagnosis (Supplementary Table S5). Notably, ASXL1 was more likely mutated in the high VAF group (61%) whereas 76% and 66% of all SF3B1 and STAG2 mutations, respectively, were identified in low VAF patients (Supplementary Table S2). High VAF patients showed a significantly shorter EFS (HR 1.77, 95%CI 1.17–2.68, p = 0.007) and OS (HR 1.67, 95%CI 1.04–2.67, p = 0.033) as compared to low VAF patients (Fig. 1B, C).

Interestingly, low VAF patients had an EFS and OS which was comparable to ELN favorable and intermediate risk patients but a significantly better EFS and OS compared to adverse risk patients (low VAF vs. favorable – EFS: HR 1.13, 95%CI 0.95–1.34, p = 0.2; OS: HR 1.07, 95%CI 0.87–1.32, p = 0.5; low VAF vs. intermediate – EFS: HR 0.96, 95%CI 0.69–1.35, p = 0.8; OS: HR 0.96, 95%CI 0.64–1.44, p = 0.8; low VAF vs. adverse risk – EFS: HR 0.68, 95%CI 0.47–0.99, p = 0.043; OS: HR 0.49, 95%CI 0.32–0.75, p < 0.001) (Supplementary Fig. S6A–F). Conversely, high VAF patients showed a significantly worse EFS and OS than both ELN favorable and intermediate risk patients (high VAF vs. favorable—EFS: HR 1.52, 95%CI 1.26–1.84, p < 0.001; OS: HR 1.38, 95%CI 1.21–1.71, p = 0.003; high VAF vs. intermediate – EFS HR 1.76, 95%CI 1.22–2.54, p = 0.003; OS: HR 1.62, 96%CI 1.07–2.45, p = 0.023) whereas the outcome was comparable to ELN adverse risk patients (EFS: HR 1.19, 95%CI 0.80–1.78, p = 0.4; OS: HR 0.83, 95%CI 0.54–1.27, p = 0.4) (Supplementary Fig. S6A–F).

After data imputation, multivariate analysis of known risk factors in AML such as age, gender and WBC count at diagnosis as well as mutations found in at least 20 patients revealed that only the VAF of MRG mutations had a significant impact both on EFS and OS (Supplementary Fig. S7).

For low VAF patients there was significant better OS when undergoing alloHCT in first complete remission (CR/CRi) (HR 0.38, 95%CI 0.25–0.56, p < 0.001). The same was seen for patients with a high VAF (HR 0.67, 95%CI 0.51–0.88, p = 0.004, Fig. 2A, B), albeit limited by small numbers. Finally, patients with a high VAF had a significantly higher cumulative incidence of relapse (CIR) compared to patients with low VAF MRG mutation (HR 3.48, 95%CI 1.48–8.16, p = 0.0042), while non relapse mortality and transplant associated characteristics were similar between these groups (HR 1.57, 95%CI 0.54–4.58, p = 0.41; Fig. 2C and Supplementary Table S6). This indicates that the more aggressive disease biology of MRG mutated AML with higher VAF is maintained even after alloHCT.

Fig. 2: Impact of type of consolidation therapy on the outcome of patients with MRG mutations.
figure 2

A Simon-Makuch plot showing the time-adjusted overall survival of all low VAF patients considering allogeneic stem cell transplantation as a time dependent-variable. 3 patients had to be excluded from the analysis due to missing data. B Same as in A but for high VAF patients. 2 patients had to be excluded from the analysis due to missing data. C Competing risk analysis of MRG low vs. high VAF patients receiving alloHCT in first CR/CRi. CIR cumulative incidence of relapse, NRM non-relapse mortality. 3 patients had to be excluded from the analysis due to missing data.

Full size image

This study provides evidence that the clone size of MRG mutations is an independent prognostic marker in AML patients with MRG mutations.

The VAF cut-off of 45% separated the patients classified as ELN adverse risk into two distinct groups with different prognosis both in uni- and multivariate analysis. Patients with a VAF below 45% had an outcome similar to ELN intermediate risk patients, whereas patients who had at least one MRG mutation in the major clone with a VAF of 45% or higher had a significantly worse outcome. As the probability to reach CR was comparable between the two groups, the worse outcome is driven by a higher risk of relapse rather than primary resistance to induction chemotherapy.

The high VAF either represents patients with heterozygous mutations, which are present in nearly all leukemic clones, homozygous mutations, or mutations in combination with loss of heterozygosity. These findings support the hypothesis that MRG mutations drive the more aggressive behavior of leukemic cells, when they are part of the dominant clone. Future studies using whole exome and single cell sequencing should evaluate if the zygosity of MRG mutations and their role within the clonal hierarchy of AML - for which the VAF is used as a surrogate - can shed further light on the prognostic impact.

Contrastingly, MRG mutations had no prognostic impact in patients with favorable cytogenetic or molecular changes, including with high or low VAF, in line with previous studies. A recent study of patients with NPM1 mutations suggested that the negative impact of MRG mutations was superseded by NPM1-measurable residual disease (MRD) after two cycles of intensive chemotherapy, emphasizing the value of MRD [10].

In the present study, a lower proportion of patients with high VAF underwent alloHCT. However, time-dependent survival analysis suggested that both groups derived benefit from undergoing alloHCT in first remission irrespective of VAF.

Taken together, our study suggests that the VAF of MRG mutations determines the prognosis of patients with these mutations. Our approach takes the clone size and thus the position of MRG mutations in the clonal evolution into account. Furthermore, our data supports the ELN recommendation to consider alloHCT for all patients bearing MRG mutations within the ELN adverse risk group.