Genetic Characteristics According to Subgroup of Acute Myeloid Leukemia with Myelodysplasia-Related Changes

Acute myeloid leukemia with myelodysplasia-related changes (AML-MRC) includes heterogeneous conditions such as previous history and specific cytogenetic and morphological properties. In this study, we analyze genetic aberrations using an RNA-based next-generation sequencing (NGS) panel assay in 45 patients with AML-MRC and detect 4 gene fusions of KMT2A-SEPT9, KMT2A-ELL, NUP98-NSD1, and RUNX1-USP42 and 81 somatic mutations. Overall, all patients had genetic aberrations comprising of not only cytogenetic changes, but also gene fusions and mutations. We also demonstrated several characteristic genetic mutations according to the AML-MRC subgroup. TP53 was the most commonly mutated gene (n = 11, 24%) and all were found in the AML-MRC subgroup with myelodysplastic syndrome-defining cytogenetic abnormalities (AML-MRC-C) (p = 0.002). These patients showed extremely poor overall survival not only in AML-MRC, but also within the AML-MRC-C subgroup. The ASXL1 (n = 9, 20%) and SRSF2 (n = 7, 16%) mutations were associated with the AML-MRC subgroup with >50% dysplasia in at least two lineages (AML-MRC-M) and were frequently co-mutated (55%, 6/11, p < 0.001). Both mutations could be used as surrogate markers to diagnose AML-MRC, especially when the assessment of multilineage dysplasia was difficult. IDH1/IDH2 (n = 13, 29%) were most commonly mutated in AML-MRC, followed by CEBPA (n = 5, 11%), PTPN11 (n = 5, 11%), FLT3 (n = 4, 9%), IDH1 (n = 4, 9%), and RUNX1 (n = 4, 9%). These mutations were not limited in any AML-MRC subgroup and could have more significance as a risk factor or susceptibility marker for target therapy in not only AML-MRC, but also other AML categories.


Introduction
The current diagnosis of acute myeloid leukemia (AML) is largely dependent on genetic aberrations [1]. In the 2016 WHO classification, gene mutations were included in the category of recurrent genetic aberrations such as NPM1 and double CEBPA mutations. Nonetheless, some AML categories are diagnosed based on bone marrow (BM) morphology and other associated findings. One of the aforementioned categories is AML with myelodysplasia-related changes (AML-MRC), which is diagnosed in patients who 2 of 10 have previous history or specific cytogenetic or morphological properties. The former two conditions are more obvious than the latter. Studies have questioned the independent predictive value of myelodysplasia in the absence of high-risk cytogenetic abnormalities in AML [2,3] and have defined multilineage dysplasia with more restrictive criteria such as micromegakaryocytes and hypogranulated neutrophils [4]. However, diagnosis remains difficult even for an experienced hematopathologist. In recent years, next-generation sequencing (NGS) has been widely used in clinics and has established the genetic characteristics and their significance in each disease category.
In this study, we analyzed the genetic aberrations in AML-MRC using an RNA-based NGS panel assay and detected gene fusions, mutations, and expressions. We compared the genetic profile among AML-MRC subgroups and endeavored to determine characteristic genetic mutations according to subgroup and to elucidate their clinical significance.

Patients and Samples
We evaluated all consecutive patients who were diagnosed with AML-MRC at Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, from 2013 to 2018. Patients' medical records, including history of myelodysplastic syndrome (MDS), myelodysplastic syndrome/myeloproliferative disorder (MDS/MPN), or chemotherapy; laboratory data, BM morphology, and immunophenotypes, were reviewed. Cytogenetic abnormalities were classified according to the 2020 International System for Human Cytogenetic Nomenclature guidelines [5]. Patients were classified into the following three subgroups: (1) patients with history of prior MDS or MDS/MPN (AML-MRC-H), (2) patients with MDS-defining cytogenetic abnormalities (AML-MRC-C), and (3) patients with >50% dysplasia in at least two lineages (AML-MRC-M). The diagnosis and classification were confirmed independently by three hematopathologists. The risk category was determined following the 2017 European LeukemiaNet (ELN) classification [6]. The study received Institutional Review Board approval from Seoul St. Mary's Hospital, The Catholic University of Korea (IRB No. KC22RISI0078).

Molecular Analysis
Anchored multiplex PCR-based enrichment RNA-sequencing libraries were generated from 250 ng of RNA using the ArcherDx FusionPlex ® Myeloid assay for Illumina (ArcherDx, Boulder, CO, USA) according to the manufacturer's instructions. Briefly, reverse transcription using random primers was performed for synthesis of cDNA, followed by end repair and adenylation steps. Cleanup of cDNA using Agencourt ® AMPure ® XP beads and ligation of molecular barcode (MBC) adapters and universal primer sites was performed. The MBC adapter-attached cDNA was amplified by the GSP1 primer pool and primer complementary to universal primer site, and a second PCR using the GSP2 primer pool was performed. The libraries were quantitated using a KAPA Universal Library Quantification Kit (Kapa Biosystems, Woburn, MA, USA), normalized, and loaded into NextSeq (Illumina). Data were analyzed by Archer ® Analysis version 5.1.7 (ArcherDX). For AMP-based NGS analysis, variants were selected and annotated using analytics algorithms and public databases [7]. The limit of detection for variant calling was set at 2%. Normalized RNA expression values were calculated by dividing the unique RNA reads for each GSP2 by the arithmetic mean of the unique RNA reads for all control GSP2s included in the panel. Relative RNA expression values were reported in RNA_expression_visualization.tsv and were visualized using heat maps. Each heatmap showed samples in columns and binned normalized per GSP2 RNA expression values (0-9) in rows.
TP53 and FLT3-internal tandem duplication mutations were separately analyzed using Sanger sequencing and fragment analysis according to methods used in previous studies, respectively [8,9]. Recurrent gene fusions were analyzed by multiplex reverse transcriptase-PCR (Bio-Rad Laboratories, Hercules, CA, USA) [10].

Statistical Analysis and Response Assessment
Categorical variables were compared using the Chi-square test or Fisher's exact test, while continuous variables were analyzed with the Mann-Whitney U test and Kruskal-Wallis H test. Overall survival (OS) curves were plotted using the Kaplan-Meier method and were analyzed with the log-rank test. Results were expressed as the hazard ratio with a 95% confidence interval (95% CI). For multivariate analysis, variables with a p-value <0.10 in the univariate analysis were entered into a Cox proportional hazards model or proportional hazards model for a subdistribution of competing risk factors. All statistical analyses were performed using SPSS, version 13.0 (SPSS, Inc., Chicago, IL, USA) and R software (version 3.4.1, R Foundation for Statistical Computing, Vienna, Austria, 2017).

Clinical Outcomes of Patients with AML-MRC Based on Disease Subgroup and Therapy
Excluding 6 patients (13.3%) without treatment, 31 (68.9%) and 8 patients (17.8%) received intensive chemotherapy and a low-intensity treatment, respectively (Table 1). In detail, among the 31 patients receiving intensive chemotherapy, 12 patients were treated with intensive chemotherapy only, while 19 patients underwent allogeneic stem cell transplantation. Among theeight8 patients receiving the low-intensity treatment, five and three patients were treated with a hypomethylating agent (HMA) and low-dose cytarabine, respectively. There were no significant differences in treatment modalities between the AML-MRC groups. When compared AML-MRC-M with the -C groups, the OS was significantly different; the estimated OS was 13.7 and 5.4 months in the AML-MRC-M and -C groups, (p = 0.004), respectively, after the median follow-up period of 84.8 months for survivors (Figure 2a). This significant survival difference was identically observed when the analysis was performed for patients receiving any treatment (p = 0.010) or only those who had undergone intensive chemotherapy (p = 0.016) (Figure 2b,c). When analyzed within the AML-MRC-C group, the TP53 mutation predicted a shorter OS not only in all patients (p = 0.006), but also in those receiving any treatment (p = 0.010) in the univariate analysis for OS (Table 2). In the multivariate analysis for OS, the AML-MRC subgroup had an independent prognostic value after adjusting for age, white blood cell count, and BM blasts at diagnosis. In addition, AML-MRC-C compared to AML-MRC-M showed a significantly worse outcome, with a threefold higher hazard ratio for death (p = 0.003) ( Table 3).  patients (p = 0.006), but also in those receiving any treatment (p = 0.010) in the univariate analysis for OS (Table 2). In the multivariate analysis for OS, the AML-MRC subgroup had an independent prognostic value after adjusting for age, white blood cell count, and BM blasts at diagnosis. In addition, AML-MRC-C compared to AML-MRC-M showed a significantly worse outcome, with a threefold higher hazard ratio for death (p = 0.003) ( Table 3).

Discussion
AML-MRC has been estimated to represent 24-35% of all AML cases and is more commonly seen in older AML patients [11]. In this study, we performed an RNA-based NGS panel assay in AML-MRC and comprehensively analyzed gene fusions, mutations, and expressions. We found four gene fusions, two with KMT2A rearrangement. The other two fusions were NUP98-NSD1 and RUNX1-USP42, which could be missed on a cytogenetic analysis. NUP98-NSD1 was reported at a relatively low frequency in AML and MDS and had an impact on poor prognosis [12,13]. The RUNX1-USP42 fusion is a rare RUNX1 rearrangement in AML and further emphasizes the need for the collection of additional cases [14]. The genetic profile was comparable to results from previous studies (Supplementary Table S3) with characteristic genetic mutations according to the AML-MRC subgroup. We also observed that the genetic profile and the prognosis of AML-MRC were not uniform, but significantly differed by the AML-MRC subgroup.
First, the TP53 mutation was most common in AML-MRC and highly associated with AML-MRC-C. All cases with the TP53 mutation had complex karyotypes, while a subset of cases with complex karyotype did not have the TP53 mutation, indicating that complex karyotypic changes generally precede the mutation [15,16]. Both AML-MRC-C and TP53 mutations were considered adverse prognostic markers, as replicated in this study. Moreover, the TP53 mutation itself was an adverse prognostic marker within AML-MRC-C. These results collectively indicated that TP53 mutations are associated with an extremely poor overall survival not only in AML-MRC, but also in the AML-MRC-C subgroup [17].
Second, ASXL1 and SRSF2 mutations were associated with AML-MRC-M, and comutations were frequent. Results from this and a previous studies showed that ASXL1/SRSF2 co-mutated AML was associated with old age, AML-MRC, and monocytic differentiation [18]. The ASXL1 mutation was defined as an adverse prognostic marker in AML [6], but we did not confirm a trend towards an inferior outcome in patients with ASXL1 or the ASXL1/SRSF2 co-mutation. We carefully postulated that ASXL1 mutated AML consists of heterogeneous cases with morphological signs of dysplasia [19], which exhibit a wide range of prognoses. In addition, the mutation of SRSF2 as well as ASXL1 could have a potential role as a surrogate marker in the AML-MRC-M subgroup, especially when the morphological assessment of multilineage dysplasia is difficult [19].
Third, IDH1 and IDH2 mutations were detected in all subgroups of AML-MRC without prognostic significance. IDH mutations were also common genetic alterations in AML and MDS. Among AML-MRC, IDH1 and IDH2 mutations have been reported at frequencies of approximately 4% and 21%, respectively [20]. A recent study revealed that the IDH1 mutation was associated with myeloid dysplasia in mice, which exhibited anemia, ineffective erythropoiesis, and increased immature progenitors and erythroblasts [21]. However, there is a considerable number of AMLs with a IDH1/IDH2 mutation categorized into groups other than AML-MRC in cases without definable multilineage dysplasia. The IDH1/IDH2 mutation has been scrutinized more as a susceptible genetic marker for current target therapies [22,23]. Taken together, these findings indicate that the presence of IDH1/IDH2 mutations is informative for therapeutic planning rather than AML categorization [24].
The last group included CEBPA, PTPN11, FLT3, and RUNX1 mutations. These mutations were detected in all AML-MRC subgroups. Previous studies demonstrated that dysplasia was not relevant for AML with CEBPA and RUNX1 mutations, and RAS pathway mutations and FLT3-ITD were significantly more frequent in cases without evaluable erythroid cells [25,26]. Accordingly, we carefully propose that these mutations could have significance as prognostic factors [27], and the classification of patients with these mutations as AML-MRC remains to be clarified. In particular, an understanding of genetic characteristics of each AML-MRC could affect treatment decisions and the therapeutic approach using a novel agent prioritized over conventional chemotherapy [28,29].
For treatment decisions, the diagnosis of AML-MRC by itself may have potential clinical implications. In a previous study by Seymour et al., authors revealed that the use of azacitidine improves clinical outcomes in older patients with AML-MRC as compared to conventional care regimens such as low-dose cytarabine [30]. Further, within [28] a similar diagnosis of AML-MRC, AML-MRC-C showed a worse independent outcome when compared to AML-MRC-M in this present study. These heterogeneities in clinical outcomes suggest that patients who are diagnosed as AML-MRC cannot be regarded as a homogeneous group, but should be considered separately based on the AML-MRC types. In particular, the understanding of each AML-MRC subtype may affect treatment decision, and the therapeutic approach using the novel agent CPX−351 could be prioritized over conventional chemotherapy for medically fit patients with AML-MRC-C and AML-MRC-H. Currently, while the approval of venetoclax in combination with azacitidine did not specifically investigate AML-MRC patients, the incidence of composite complete remission with this combination versus azacitidine alone was notably improved across all AML risk groups, including adverse cytogenetic risk and secondary AML [31], which might suggest potential benefit in the AML-MRC subgroup [32].
Although we clearly demonstrated the characteristic genetic changes in AML-MRC, several limitations were noted. We did not fully evaluate MDS-associated genetic mutations because we used an RNA-based NGS panel assay [33]. We did not include other AML categories whose diagnoses were based on morphology. Issues of total sample size and a comprehensive analysis linked with gene expression need to be addressed in further studies.
In summary, AML-MRC is composed of heterogenous cases with different risk categories and genetic characteristics. Most AML-MRC patients had genetic aberrations, including gene fusions and mutations, as well as cytogenetic changes. In terms of genetic mutations, AML-MRC showed characteristics according to subgroup, and each had its own significance. The TP53 mutation was closely associated with AML-MRC-C and the extremely poor outcome in AML-MRC. ASXL1 and SRSF2 mutations were associated with AML-MRC-M and could be used as surrogate markers to diagnose AML-MRC. Mutations in IDH1/IDH2, CEBPA, PTPN11, FLT3, and RUNX1 were not limited to any AML-MRC subgroup and could be more significant as risk factors or susceptible markers for target therapy in not only AML-MRC, but also other AML categories.