Introduction

The hypomethylating agents (HMA) decitabine (DAC) and azacitidine (AZA) are a standard of care in AML and higher risk MDS patients not eligible for intensive treatment. While dynamic features, such as early platelet response [1], can be used to estimate eventual treatment response, no pre-treatment markers are in routine clinical use.

Several studies reported associations between outcomes of patients with MDS or AML receiving HMAs and genetic aberrations, DNA methylation, mRNA or microRNA expression, or other markers (e.g., HbF) [2,3,4,5,6,7,8,9,10,11,12,13,3).

Table 3 Characteristics of patients with samples analyzed via panel sequencing according to the TP53 mutation status

The TP53 mutations were all present in the major AML clone of the respective patient. Only one patient (13%) with a TP53-mutated AML had a minor subclone, while 8 (32%) of the patients with TP53 wild-type AML did. Patients with a TP53 mutation harbored a median of only one additional mutation (range, 0–4).

Association of TP53 mutations with clinical features and outcome

Compared with TP53 wild type, patients with TP53 mutations were younger (P = 0.01; median, 71 vs 77 years); AML with TP53 mutation tended to more often develop from antecedent MDS (P = 0.12) (Table 3).

In the outcome comparisons between AML patients with mutated or wild-type TP53, there were no differences in the response rates, but patients with TP53 mutations had a shorter OS than those with wild-type TP53 (P = 0.036) (Table 2, Fig. 1d).

Twenty-four of the patients with sequenced samples had received ATRA. Of these, only three harbored a TP53 mutation, which precluded outcome analyses.

Association between other mutations and outcome

Other markers previously reported to be associated with outcomes in DAC-treated patients are mutations in SRSF2, [11] DNMT3A, [2, 17] TET2, [7] IDH1, or IDH2. [8, 14] We found no differences in response rates or OS between patients with or without mutations in these genes or in at least one RNA splicing gene (data not shown). Moreover, there were no differences in response rates and OS between patients with ≤ 3 mutated genes and those with > 3 mutated genes (Supplemental Table S3).

However, the presence of subclones was associated with a shorter OS in comparison with their absence (P = 0.05) (Fig. 1e, Supplemental Table S3). Only one of the 9 patients with a minor subclone also harbored a TP53 mutation. Since both a TP53 mutation and presence of a minor subclone were associated with shorter OS, we combined patients with at least one of these features into one group and compared them with the remaining. Compared with patients with TP53 wild type and absence of a minor subclone, those with a TP53 mutation or a minor subclone expectedly had shorter OS (P = 0.01); no differences in the response rates were observed (Fig. 1f, Supplemental Table S3).

Discussion

HMAs have become a standard of care in AML patients not eligible for intensive chemotherapy. Chromosomal or molecular aberrations of TP53 are likely central in the investigation of markers and biological pathways associated with the response to HMAs [5, 7, 10,11,12, 14, 16, 18, 19]. Thus, we sought to investigate the impact of loss of 17p and TP53 mutations in our phase II trial 00331 in which older unfit AML patients were treated with 3-day DAC.

We observed that patients with a loss of 17p tended to have higher rates for CR/PR/ALE both in analyses including all patients and in those restricted to CK+ and MK+ patients. Among CK+ and MK+ patients, patients with loss of 17p also had longer median OS, but this favorable course could not be maintained over time. Patients with TP53-mutated AML had similar rates of CR/PR/ALE but shorter OS than those with wild-type TP53 (P = 0.036).

Published data on the impact of chromosome 17p aberrations on response to HMA treatment are scarce. Nazha et al. [20] observed no difference in the response rates according to chromosome 17 aberrations in MK+ and CK+ patients treated with HMAs. In an explorative retrospective analysis of the AZA-AML-001 study, patients with chromosome 17p aberrations had a strong trend for better OS when treated with AZA as compared with conventional care regimens (mainly low-dose cytarabine) [16].

The impact of TP53 mutations (or expression) on outcomes in HMA-treated MDS or AML patients has been assessed in several studies and yielded heterogeneous results. [7, 10,11,12, 14, 16, 19] Welch et al. [11] observed in patients with AML or MDS that achievement of CR was more frequent in patients with a TP53 mutation. Moreover, in contrast to the poor OS of TP53-mutated AML patients after standard induction, there was no OS difference according to TP53 mutation status in patients receiving DAC. In the aforementioned analysis of the AZA-AML-001 study, patients with TP53-mutated AML had strong trends for improved OS when treated with AZA compared with alternative therapies [16]. However, in studies among MDS patients treated with DAC or AZA, TP53 mutations had no impact on response rates but were associated with shorter response duration and/or OS [7, 10]. In another report on patients with MDS (mostly with blast excess) who received DAC, most patients with TP53 mutations achieved a CR, but they still had inferior OS [14].

In MDS, mono-allelic TP53 mutations associate with more favorable disease features (including less frequent complex karyotype and better OS) than multi-hit TP53 mutations [21]. The role of TP53 mutations under consideration of their allelic state remains to be established. Welch et al. [11] described that two-thirds of the patients with TP53 mutations potentially had both alleles affected. In our study, 5 patients fulfilled the criteria of a TP53 multi-hit mutation (i.e., multiple gene mutations or gene mutation plus genomic loss) according to Bernard et al. [21]. The low patient numbers precluded meaningful outcome analyses. Hopefully, future studies will be able to decipher the impact of the allelic state of TP53 and its impact in response to DAC.

Considering our results and those reported by others, it currently remains elusive whether the TP53 aberrations or rather associated genetic features may confer sensitivity to HMAs. As observed in the present study, TP53 mutations and chromosome 17p aberrations coincide with CK+ and MK+ [22,23,24,25,26]; and for trial 00331, which was subject to the present study, we previously reported that MK+ patients had higher response rates and similar OS compared with MK− patients [3]. Similar results have been reported by Wierzbowska et al. [27] from a post hoc analysis of the phase 3 DACO-016 trial in AML and by our group from the phase 3 EORTC trial 06011 in MDS [9, 28]. In the study by Welch et al. [11], almost all patients with TP53 mutations had unfavorable cytogenetics, and achieving a CR was more frequent in patients with unfavorable than those with intermediate or favorable cytogenetics. In the study by Chang et al. [14], almost all TP53-mutated patients who achieved a CR were CK+ or had monosomies.

Welch et al. [11] suggested that a variable response of TP53-mutated AML to DAC may be due to the presence of TP53 mutations in subclones instead of the major clone. We observed no superior response to DAC, although TP53 mutations were all present in the major clone, and despite other features supporting their disease-driving effect, i.e. TP53-mutated AML only rarely harbored a minor subclone and had a low number of additional mutations [7]. However, we did observe that patients with a minor subclone had shorter OS than those without, although only one of the nine patients with a minor subclone also harbored a TP53 mutation.

The heterogeneity in the reports on associations between TP53 aberrations and response to HMAs may be due to weaknesses that are variably shared by the studies, including the present study. First, analyses are often based on relatively small patient numbers [11, 14, 16]. Second, if provided, the information on 17p loss normally stems from conventional cytogenetics, although the (presumably lost) TP53 allele may be present in unidentified chromosome material. [16, 29] Third, cohorts variably comprise MDS or AML patients or both, although DAC may have higher efficacy in patients with higher blast counts [30, 31]. Fourth, there is the heterogeneity in treatment. In several studies, patients treated with DAC or AZA were combined into one group [7, 10], or patients were included who received DAC combined with another agent [2, 3, 7, 10]. Moreover, DAC was administered according to different protocols. The majority of the patients received DAC according to the 5-day protocol (total of 100 mg/m2 over 5 days) [10, 14]. In the study by Welch et al. [11], the majority of patients received DAC according to the 10-day protocol (total 200 mg/m2 over 10 days). Patients in our present study received DAC according to the 3-day protocol (total of 135 mg/m2 over 3 days; every 6 weeks), in part of the patients followed by a reduced dosage maintenance phase. Moreover, in the present study, the patients with loss of 17p or TP53 mutation had received only a median of 2 (range, 1–12) or 1 (range, 1–6) DAC courses, while several courses are normally required to achieve best response.

While the clinical observation of the (counter-intuitive) response to HMAs in adverse genetics AML/MDS is more and more accepted within the clinical community, the underlying mechanism of the interaction between hypomethylating activity and these genotypes is still unresolved. Monosomal chromosomal regions may preferentially attract epigenetic silencing [32, 33], providing a particularly sensitive target to DNMT inhibition. Despite the present lack of a conclusive model of this interaction, clinicians need to be aware that the responses, while surprisingly frequent, are often short-lived. Hence, they can also be quite deceptive, by raising unfounded optimism regarding their duration. Thus, patients with adverse genetics who are eligible for allografting should transition to this curative treatment in a timely manner, i.e., before HMA resistance sets in.

In summary, within the specifications of the patient cohort studied, loss of 17p was associated with trends for higher DAC response rates, both within the entire cohort and among patients with CK+ or MK+ AML. Patients with a TP53 mutation achieved similar response rates as patients with wild-type TP53, but had a shorter OS. Our data further support the potential applicability of TP53 aberrations as predictor for HMA treatment, and emphasize a possible role for subclonal mutations in this regard. Isolated TP53 mutation analyses apparently are not sufficient for prediction of HMA response. Cytogenetic analysis remains standard and allows for evaluation of MK+ status and 17p loss. The landscape of HMA-based therapy is changing, and favorable responses in adverse genetics patients are also observed when these drugs are combined with the BCL-2 inhibitor venetoclax [34] or all-trans retinoic acid [35]. Hence, the impact of the different types of adverse cytogenetics and TP53 alterations (e.g., cytogenetic and molecular genetic mono- or bi-allelic loss) on outcome after HMA combination studies will be of great interest.