Abstract
To explore the distinct mutation profiles between acute myeloid leukemia with myelodysplasia-related changes (AML-MRC) and de novo AML and their relationships with prognosis.
Next-generation sequencing of 42 myeloid neoplasm-related genes in 293 newly diagnosed patients with AML.
Eighty-four patients had AML-MRC, and 161 patients had de novo AML. The mutation rates of ASXL1 (25% vs 8.7%, P = .001), NRAS (17.9% vs 8.1%, P = .022), PTPN11 (11.9% vs 5%, P = .048), SETBP1 (6% vs 0.6%, P = .033), SRSF2 (11.9% vs 5.6%, P = .08), TP53 (16.7% vs 1.2%, P < .001), and U2AF1 (17.9% vs 7.5%, P = .014) in AML-MRC were higher than those in de novo AML, while the rates of FLT3-ITD (3.6% vs 15.5%, P = .005), KIT (0% vs 6.2%, P = .046), WT1 (3.6% vs 9.9%, P = .077), NPM1 (1.2% vs 21.7%, P < .001), and CEBPA (4.8% vs 24.2%, P < .001) mutation were lower. The appearance of ASXL1, TP53, U2AF1, SRSF2, and SETBP1 mutation could predict AML-MRC–like features in de novo AML, which was related to older age (60 vs 51 years, P = .001), low WBC counts (4.7 × 109/L vs 11.6 × 109/L, P = .001), and inferior outcomes (median overall survival, 15 months vs not reached, P = .003).
The presence of AML-MRC–related mutations can reveal a subset of patients with de novo AML similar to patients with AML-MRC.
The diagnostic criteria of acute myeloid leukemia with myelodysplasia-related changes (AML-MRC) can be influenced by many factors, such as bone marrow sampling site, blast cell count, and hematopathologist experience.
AML-MRC and de novo AML exhibit distinguishing mutation profiles. ASXL1, NRAS, PTPN11, SETBP1, SRSF2, TP53, and U2AF1 are typically mutated in AML-MRC while FLT3-ITD, KIT, WT1, NPM1, and CEBPA are not.
AML-MRC–related mutations can help distinguish AML patients with AML-MRC–like features.
INTRODUCTION
Acute myeloid leukemia (AML) is a heterogenous clonal myeloid neoplasm that can develop from a preleukemia condition or arise de novo. AML with myelodysplasia-related changes (AML-MRC) is a subgroup of AML featuring older age, lower blast counts, lower remission rate, and shorter overall survival (OS) than de novo AML.1 Patients with AML-MRC exhibit myelodysplastic syndrome (MDS)–related features, including a history of MDS or MDS/myeloproliferative neoplasm (MDS/MPN) history, MDS-related cytogenetic abnormalities, and multilineage dysplasia (MLD) without NPM1 or biCEBPA mutations.2 Considering their poor outcomes, developing adaptive treatment strategies in advance is important. For example, CPX-351 is currently the only drug approved by the US Food and Drug Administration for the treatment of adults with newly diagnosed AML-MRC because of its higher rates of complete remission (CR) or CR with incomplete hematologic recovery (CRi) and prolonged OS in AML-MRC vs 7 + 3 therapy.3 However, the diagnostic cutoff of 20% blast cells used to differentiate between AML and MDS is artificial and may be influenced by sampling sites or other factors when the blast cell count is near the diagnostic cutoff. Previous studies have found that high-risk MDS and AML with low blast counts not only share similar biological features, including bone marrow biomarkers,4 frequent poor-risk cytogenetics, and mutation profiles,5 but also similar clinical features, such as old age, low WBC count, and comparable poor prognosis.6 Therefore, differentiating between AML and MDS solely based on blast count diagnostic criteria may lead to a bias in the judgment of MDS history. In addition, MLD should be confirmed by experienced hematopathologists with aspirate samples with sufficient hematopoietic precursors, which can be difficult to accomplish in the clinic. Therefore, MLD can sometimes be mistaken for a lack of hematopoietic precursors. With the advent of next-generation sequencing (NGS), the presence of some mutations in multiple cell lineages has been identified as a potential distinguishing feature of secondary AML (sAML).7,8 Therefore, NGS provides us with a possible way to identify patients with probable AML-MRC when the MDS history and MLD are unclear.
In this study, we compared mutation landscapes in a Chinese cohort of 293 newly diagnosed patients with AML. Of these patients, 84 had AML-MRC, and 161 had de novo AML. We further evaluated the therapeutic outcomes associated with and prognostic impacts of high-frequency mutated genes in AML-MRC. We hypothesize that mutations specific for AML-MRC can predict AML-MRC–like features, including old age, lower blast count and WBC count, and inferior survival, even in de novo AML.
MATERIALS AND METHODS
Patients
We retrospectively evaluated all patients who underwent mutational analysis of bone marrow aspirate or peripheral blood samples by NGS and were newly diagnosed with AML but not acute promyelocytic leukemia with PML-RARA at Peking Union Medical College Hospital from January 2016 to January 2021. Informed consent was obtained in accordance with the Declaration of Helsinki. The diagnosis of AML was achieved by morphologic, cytochemical, and immunohistochemical analysis according to World Health Organization (WHO) 2016 criteria.2 Karyotyping was performed on fresh bone marrow samples by standard procedures, and karyotype was reported according to the International System for Human Cytogenetic Nomenclature 2013 nomenclature.9 Patients without previous cytotoxic or radiation therapy who did not meet any criteria for AML-MRC or had recurrent cytogenetic abnormalities were categorized as having de novo AML. Risk stratification was performed according to the 2017 European Leukemia Net classification (ELN-2017).10 Standard-intensity induction therapy included 7 days of cytarabine at 100 mg/m2 or more daily plus 3 days of daunorubicin at 45 mg/m2 or more or idarubicin at 10 mg/m2 or more daily. CR or CRi was defined as bone marrow blasts less than 5% and an absence of circulating blasts and extramedullary disease.10
Gene Sequencing and Analysis
Genomic DNA was extracted from bone marrow or peripheral blood samples and sent for polymerase chain reaction analysis and sequencing. The panel of 42 myeloid neoplasm-related genes we used here encompassed spliceosome (U2AF1, SRSF2, SF3B1, and ZRSR2), DNA methylation (DNMT3A, TET2, IDH2, and IDH1), kinase signaling (NRAS, PTPN11, KRAS, JAK2, SETBP1, FLT3, NF1, STAT3, CSF3R, CBL, KIT, and PDGFRA), transcription factor (RUNX1, GATA2, ETV6, and CEBPA), cohesion (STAG2, SMC3, SMC1A, and RAD21), chromatin structure (ASXL1, EZH2, KMT2A, PHF6, BCOR, and BCORL1), and other genes (TP53, NPM1, WT1, CALR, PPM1D, SH2B3, MPL, and PIGA). DNA libraries were sequenced on an Illumina HiSeq platform, and raw data were filtered by FastQC. Clean data were mapped to the reference human genome (UCSC hg38)11 by a Burrows-Wheeler aligner.12 Single-nucleotide polymorphisms, insertions, and deletions detected by the Genome Analysis Toolkit (GATK) were annotated by ANNOVAR13 according to related existing databases. Mutations reported in COSMIC as hematologic malignancy associated were all reserved. Any genetic abnormality was removed when fulfilling any of the following conditions: (1) sequencing depth of less than 200×, (2) variant allele frequency less than 2% or more than 90%, (3) mention in the 1000 Genomes Project or ExAC with a minor allele frequency over 1%, (4) categorization as tolerated or benign mutation in both SIFT and FATHMM, and (5) ClinVar categorization as benign or likely benign.
Statistical Analysis
Data were analyzed with SPSS 26.0 software (SPSS) and R software (version 4.0.3; https://www.r-project.org/). Student t test or the Mann-Whitney U test was applied to analyze group differences for continuous variables, and Pearson χ 2 test was applied to assess group differences in categorical variables. OS was calculated as the time from the date of diagnosis to death or last follow-up. OS curves were plotted using the Kaplan-Meier method, compared using a log-rank test, and analyzed by a univariate Cox model. The significance of the hazard ratio (HR) was assessed by the Wald test.
RESULTS
Patients’ Characteristics
A total of 293 patients with AML were included in this analysis. Eighty-four (28.7%) patients met the diagnostic criteria for AML-MRC. Among them, 46 (54.8%) carried MDS-related cytogenetic changes, 48 (57.1%) had a history of MDS or MDS/MPN, and 29 (34.5%) had MLD (Supplementary Figure 1; all supplemental materials can be found at American Journal of Clinical Pathology online). Regarding the patients with MDS-related cytogenetic changes, 34 (73.9%) patients had complex karyotypes, 8 (17.4%) patients had –7/del(7q), 3 (6.5%) patients had –5/del(5q), and 1 (2.2%) patient had i(17q)/t(17p), while 16 (19%) had normal karyotypes. A total of 161 (54.9%) patients were classified as having de novo AML, and 84 (52.2%) of them exhibited a normal karyotype. Forty-two (14.3%) of all patients with AML could not be categorized due to a lack of karyotype information, and 6 (2%) patients had therapy-related AML. The patients’ baseline characteristics are summarized in Table 1.
Baseline Characteristics of Patients With AML-MRC and De Novo AML
| Characteristic | AML-MRC (n = 84) | De Novo AML (n = 161) | P Value |
|---|---|---|---|
| Sex, male, No. (%) | 51 (61.5) | 99 (60.7) | .9 |
| Age, median (range), y | 62 (22-80) | 52 (14-88) | <.001 |
| Bone marrow blasts, median (range), % | 39 (20-96) | 58 (10-98) | <.001 |
| WBCs, median (range), × 109/L | 5.0 (0.3-294.6) | 9.4 (0.4-362.6) | .028a |
| Hemoglobin, median (range), g/L | 76 (29-132) | 80 (42-159) | .51 |
| Platelets, median (range), × 1012/L | 48.5 (2-766) | 39.0 (3-669) | .22 |
| Mutated genes, mean (range), No. | 2.6 (0-8) | 2.3 (0-8) | .19 |
| Karyotype, No. (%) | |||
| Complex karyotype | 34 (40.5) | 0 | — |
| –7/del(7q) | 8 (9.5) | 0 | — |
| –5/del(5q) | 3 (3.6) | 0 | — |
| i(17q)/t(17p) | 1 (1.2) | 0 | — |
| Normal karyotype | 16 (19) | 85 (52.8) | — |
| t(8;21)(q22;q22) | 0 | 11 (6.8) | — |
| inv(16)(p13.1q22)/t(16;16)(p13.1;q22) | 0 | 21 (13) | — |
| ELN-2017 subgroup, No. (%) | |||
| Favorable | 0 (0) | 47 (29.2) | — |
| Intermediate | 17 (20.2) | 85 (52.8) | — |
| Adverse | 57 (67.9) | 29 (18) | — |
| Not classified | 10 (11.9) | 0 (0) | — |
| Characteristic | AML-MRC (n = 84) | De Novo AML (n = 161) | P Value |
|---|---|---|---|
| Sex, male, No. (%) | 51 (61.5) | 99 (60.7) | .9 |
| Age, median (range), y | 62 (22-80) | 52 (14-88) | <.001 |
| Bone marrow blasts, median (range), % | 39 (20-96) | 58 (10-98) | <.001 |
| WBCs, median (range), × 109/L | 5.0 (0.3-294.6) | 9.4 (0.4-362.6) | .028a |
| Hemoglobin, median (range), g/L | 76 (29-132) | 80 (42-159) | .51 |
| Platelets, median (range), × 1012/L | 48.5 (2-766) | 39.0 (3-669) | .22 |
| Mutated genes, mean (range), No. | 2.6 (0-8) | 2.3 (0-8) | .19 |
| Karyotype, No. (%) | |||
| Complex karyotype | 34 (40.5) | 0 | — |
| –7/del(7q) | 8 (9.5) | 0 | — |
| –5/del(5q) | 3 (3.6) | 0 | — |
| i(17q)/t(17p) | 1 (1.2) | 0 | — |
| Normal karyotype | 16 (19) | 85 (52.8) | — |
| t(8;21)(q22;q22) | 0 | 11 (6.8) | — |
| inv(16)(p13.1q22)/t(16;16)(p13.1;q22) | 0 | 21 (13) | — |
| ELN-2017 subgroup, No. (%) | |||
| Favorable | 0 (0) | 47 (29.2) | — |
| Intermediate | 17 (20.2) | 85 (52.8) | — |
| Adverse | 57 (67.9) | 29 (18) | — |
| Not classified | 10 (11.9) | 0 (0) | — |
AML, acute myeloid leukemia; AML-MRC, acute myeloid leukemia with myelodysplasia-related changes; ELN-2017, 2017 European Leukemia Net.
Baseline Characteristics of Patients With AML-MRC and De Novo AML
| Characteristic | AML-MRC (n = 84) | De Novo AML (n = 161) | P Value |
|---|---|---|---|
| Sex, male, No. (%) | 51 (61.5) | 99 (60.7) | .9 |
| Age, median (range), y | 62 (22-80) | 52 (14-88) | <.001 |
| Bone marrow blasts, median (range), % | 39 (20-96) | 58 (10-98) | <.001 |
| WBCs, median (range), × 109/L | 5.0 (0.3-294.6) | 9.4 (0.4-362.6) | .028a |
| Hemoglobin, median (range), g/L | 76 (29-132) | 80 (42-159) | .51 |
| Platelets, median (range), × 1012/L | 48.5 (2-766) | 39.0 (3-669) | .22 |
| Mutated genes, mean (range), No. | 2.6 (0-8) | 2.3 (0-8) | .19 |
| Karyotype, No. (%) | |||
| Complex karyotype | 34 (40.5) | 0 | — |
| –7/del(7q) | 8 (9.5) | 0 | — |
| –5/del(5q) | 3 (3.6) | 0 | — |
| i(17q)/t(17p) | 1 (1.2) | 0 | — |
| Normal karyotype | 16 (19) | 85 (52.8) | — |
| t(8;21)(q22;q22) | 0 | 11 (6.8) | — |
| inv(16)(p13.1q22)/t(16;16)(p13.1;q22) | 0 | 21 (13) | — |
| ELN-2017 subgroup, No. (%) | |||
| Favorable | 0 (0) | 47 (29.2) | — |
| Intermediate | 17 (20.2) | 85 (52.8) | — |
| Adverse | 57 (67.9) | 29 (18) | — |
| Not classified | 10 (11.9) | 0 (0) | — |
| Characteristic | AML-MRC (n = 84) | De Novo AML (n = 161) | P Value |
|---|---|---|---|
| Sex, male, No. (%) | 51 (61.5) | 99 (60.7) | .9 |
| Age, median (range), y | 62 (22-80) | 52 (14-88) | <.001 |
| Bone marrow blasts, median (range), % | 39 (20-96) | 58 (10-98) | <.001 |
| WBCs, median (range), × 109/L | 5.0 (0.3-294.6) | 9.4 (0.4-362.6) | .028a |
| Hemoglobin, median (range), g/L | 76 (29-132) | 80 (42-159) | .51 |
| Platelets, median (range), × 1012/L | 48.5 (2-766) | 39.0 (3-669) | .22 |
| Mutated genes, mean (range), No. | 2.6 (0-8) | 2.3 (0-8) | .19 |
| Karyotype, No. (%) | |||
| Complex karyotype | 34 (40.5) | 0 | — |
| –7/del(7q) | 8 (9.5) | 0 | — |
| –5/del(5q) | 3 (3.6) | 0 | — |
| i(17q)/t(17p) | 1 (1.2) | 0 | — |
| Normal karyotype | 16 (19) | 85 (52.8) | — |
| t(8;21)(q22;q22) | 0 | 11 (6.8) | — |
| inv(16)(p13.1q22)/t(16;16)(p13.1;q22) | 0 | 21 (13) | — |
| ELN-2017 subgroup, No. (%) | |||
| Favorable | 0 (0) | 47 (29.2) | — |
| Intermediate | 17 (20.2) | 85 (52.8) | — |
| Adverse | 57 (67.9) | 29 (18) | — |
| Not classified | 10 (11.9) | 0 (0) | — |
AML, acute myeloid leukemia; AML-MRC, acute myeloid leukemia with myelodysplasia-related changes; ELN-2017, 2017 European Leukemia Net.
Compared with patients with de novo AML, patients with AML-MRC were older at diagnosis (62 vs 52 years, P < .001) and had a lower percentage of bone marrow blasts (39.3% vs 57.5%, P < .001). The median WBC count in peripheral blood in patients with AML-MRC was lower than that in patients with de novo AML (5.0 × 109/L vs 9.37 × 109/L, P = .028), but there was no difference in the hemoglobin level or platelet count. Overall, 57 (67.9%) patients with AML-MRC were categorized into the ELN-2017 adverse group, and 17 (20.2%) were categorized into the ELN-2017 intermediate group, with no patients with AML-MRC in the ELN-2017 favorable group. Risk stratification was not performed for 10 patients with AML-MRC due to a lack of karyotype information, but all of them met either the MDS history or MLD criteria and lacked recurrent cytogenetic abnormalities according to results of fluorescence in situ hybridization analysis or sequencing of fusion genes. More than half of the patients with de novo AML were intermediate risk (85, 52.8%), while 47 (29.2%) and 29 (18%) patients were in the favorable and adverse subgroups, respectively.
Mutation Landscapes in AML-MRC and De Novo AML
Mutation heatmaps of AML-MRC and de novo AML are presented in Figure 1. The mutation frequencies of 42 genes and the differences in mutation frequencies between the two AML groups are shown in Figure 2. Eighty-two (92.1%) of 84 patients with AML-MRC, 142 (88.2%) of 161 patients with de novo AML, and 262 (89.1%) of 294 patients with AML had at least one mutation detected. AML-MRC and de novo AML exhibited distinct mutation profiles. The mutation rates of ASXL1 (25% vs 8.7%, P = .001), NRAS (17.9% vs 8.1%, P = .022), PTPN11 (11.9% vs 5%, P = .048), SETBP1 (6% vs 0.6%, P = .033), SRSF2 (11.9% vs 5.6%, P = .08), TP53 (16.7% vs 1.2%, P < .001), and U2AF1 (17.9% vs 7.5%, P = .014) were higher in AML-MRC than those in de novo AML. The mutation rates of FLT3-ITD (3.6% vs 15.5%, P = .005), KIT (0% vs 6.2%, P = .046), WT1 (3.6% vs 9.9%, P = .077), NPM1 (1.2% vs 21.7%, P < .001), and CEBPA (4.8% vs 24.2%, P < .001) were higher in de novo AML than those in AML-MRC. Detailed information about the gene mutation frequencies in the two AML groups is shown in Supplementary Table 1.
Mutation heatmaps in acute myeloid leukemia with myelodysplasia-related changes (A) and de novo acute myeloid leukemia (B).
A, Mutation frequencies of 42 genes in acute myeloid leukemia with myelodysplasia-related changes (AML-MRC) (n = 84) (left) and de novo acute myeloid leukemia (AML) (n = 161) (right). B, Differences in mutations between AML-MRC (n = 84) and de novo AML (n = 161).
We next assessed the patterns of co-occurring and mutually exclusive mutations in genes related to AML-MRC. Co-occurring gene mutations in all patients with AML are shown in Figure 3A, and Figure 3B displays the odds ratios and P values of co-occurring and mutually exclusive mutations. ASXL1 tended to be comutated with splicing factor genes, including SETBP1 (P < .001), SRSF2 (P < .001), and U2AF1 (P < .001), as well as other genes including STAG2 (P < .001), CSFR3 (P = .038), and ETV6 (P = .017), but this mutation was mutually exclusive with FLT3-ITD (P = .022), NPM1 (P = .003), and WT1 (P = .099) mutations. SETBP1 mutations tended to co-occur with SRSF2 (P = .001), NRAS (P = .058), and ETV6 (P = .038) mutations. SRSF2 mutations tended to co-occur with IDH2 (P = .028), RUNX1 (P = .004), and STAG2 (P = .011) mutations. The splicing factor gene U2AF1 tended to be comutated with ETV6 (P = .001) and IDH1 (P = .045) but never with NPM1 (P = .033). TP53 mutations never co-occurred with CEBPA (P = .077) or NPM1 (P = .082) mutations. For de novo AML, WT1 mutations tended to co-occur with FLT3-ITD (P = .001) and CEBPA (P < .001) mutations. In addition, NPM1 mutation was closely related to FLT3-ITD mutation (P < .001). Therefore, some AML-MRC–related gene mutations had a tendency to co-occur with each other and sAML-related mutations reported by Lindsley et al,7 such as mutations in STAG2, but were mutually exclusive with mutations specific to de novo AML (NPM1, CEBPA, and FLT3-ITD). Detailed mutation patterns are shown in Supplementary Table 2. These results illustrate that the mutation profiles of AML-MRC and de novo AML are different and can be considered separate entities.
Co-occurring and mutually exclusive mutations in all patients with acute myeloid leukemia (AML). A, Circos plots revealing the frequencies of co-occurring genetic alterations in all patients with AML. B, Odds ratios (ORs) and P values of comutations and mutually exclusive mutations.
Prognostic Value of Mutations Specific to AML-MRC
We further evaluated the prognostic value of unique mutations with high frequencies in AML-MRC (ASXL1, NRAS, PTPN11, SETBP1, SRSF2, TP53, and U2AF1 mutations). All patients with AML (n = 293) were included in the following prognostic analysis. The median follow-up time for all patients with AML was 18.9 months (range, 0.03-48.9 months). The OS of AML-MRC was significantly shorter than that of de novo AML (7.8 vs 40.4 months, P < .001). Patients carrying the TP53 mutation had a significantly shorter OS than those without such mutations (3.7 vs 19 months, P < .001), as did patients with SETBP1 mutations (2.4 vs 17.8 months, P < .001) and SRSF2 mutations (5.2 vs 19 months, P = .007). Carrying an ASXL1 mutation (10.9 vs 17.7 months, P = .13) or a U2AF1 mutation (10.6 vs 17.7 months, P = .07) predicted shorter median OS, but P values were not significant. Considering previous reports about the adverse survival effects of ASXL1 and U2AF1 mutations14-16 and incorporation of ASXL1 mutation into ELN-2017 adverse stratification,10 we suspect that ASXL1 and U2AF1 mutations are inferior prognostic markers. No difference in median OS was observed between AML with and without NRAS mutation (25.3 vs 16.5 months, P = .37) or between AML with and without PTPN11 mutation (14.8 vs 17.7 months, P = .99), so NRAS and PTPN11 seem to show high mutation rates in AML-MRC but do not portend poor prognosis. The OS curves mentioned above are shown in Supplementary Figure 1. Therefore, we wondered whether the presence of ASXL1, TP53, U2AF1, SETBP1, and SRSF2 mutations, which showed high frequencies in AML-MRC and conveyed inferior OS, can predict AML-MRC–like features in de novo AML.
Prediction of AML-MRC–Like Features According to AML-MRC–Related Mutations in De Novo AML
Twenty-eight (17.4%) of 161 patients with de novo AML had at least one mutation in ASXL1, TP53, U2AF1, SETBP1, or SRSF2. The median age of these 28 patients was older than that of the remaining patients with de novo AML (60 vs 51 years, P = .001), and they also had lower WBC counts (4.7 × 109/L vs 11.6 × 109/L, P = .001). The details are shown in Table 2.
Baseline Characteristics of All Patients With De Novo AML and De Novo AML With or Without MRC-Related Mutations (Mutations in ASXL1, TP53, U2AF1, SRSF2 and SETBP1)
| Characteristic | De Novo AML (n = 161) | De Novo AML With MRC-Related Mutations (n = 28) | De Novo AML Without MRC-Related Mutations (n = 133) | P Value |
|---|---|---|---|---|
| Sex, male, No. (%) | 99 (60.7) | 20 (71.4) | 79 (59.4) | .23 |
| Age, median (range), y | 52 (14-88) | 60 (27-85) | 51 (14-88) | .001 |
| Bone marrow blasts, median (range), % | 58 (10-98) | 56 (13.5-94.5) | 58 (10-98) | .52 |
| WBCs, median (range), × 109/L | 9.4 (0.4-362.6) | 4.7 (0.36-94.16) | 11.6 (0.5-362.6) | .001 |
| Hemoglobin, median (range), g/L | 80 (42-159) | 79 (53-116) | 80 (42-159) | .43 |
| Platelets, median (range), × 1012/L | 39 (3-669) | 31 (5-196) | 42 (3-669) | .33 |
| ELN-2017 subgroup, No. (%) | ||||
| Favorable | 0 | 4 (14.3) | 43 (32.3) | — |
| Intermediate | 17 (10.6) | 11 (39.3) | 74 (55.6) | — |
| Adverse | 57 (35.4) | 13 (46.4) | 16 (12) | — |
| Not classified | 10 (6.2) | 0 | 0 (0) | — |
| Karyotype, No. (%) | ||||
| Normal karyotype | 85 (52.8) | 10 (35.7) | 75 (56.4) | — |
| t(8;21)(q22;q22) | 11 (6.8) | 3 (10.7) | 8 (6) | — |
| inv(16)(p13.1q22)/t(16;16)(p13.1;q22) | 21 (13) | 16 (57.1) | 5 (3.8) | — |
| Characteristic | De Novo AML (n = 161) | De Novo AML With MRC-Related Mutations (n = 28) | De Novo AML Without MRC-Related Mutations (n = 133) | P Value |
|---|---|---|---|---|
| Sex, male, No. (%) | 99 (60.7) | 20 (71.4) | 79 (59.4) | .23 |
| Age, median (range), y | 52 (14-88) | 60 (27-85) | 51 (14-88) | .001 |
| Bone marrow blasts, median (range), % | 58 (10-98) | 56 (13.5-94.5) | 58 (10-98) | .52 |
| WBCs, median (range), × 109/L | 9.4 (0.4-362.6) | 4.7 (0.36-94.16) | 11.6 (0.5-362.6) | .001 |
| Hemoglobin, median (range), g/L | 80 (42-159) | 79 (53-116) | 80 (42-159) | .43 |
| Platelets, median (range), × 1012/L | 39 (3-669) | 31 (5-196) | 42 (3-669) | .33 |
| ELN-2017 subgroup, No. (%) | ||||
| Favorable | 0 | 4 (14.3) | 43 (32.3) | — |
| Intermediate | 17 (10.6) | 11 (39.3) | 74 (55.6) | — |
| Adverse | 57 (35.4) | 13 (46.4) | 16 (12) | — |
| Not classified | 10 (6.2) | 0 | 0 (0) | — |
| Karyotype, No. (%) | ||||
| Normal karyotype | 85 (52.8) | 10 (35.7) | 75 (56.4) | — |
| t(8;21)(q22;q22) | 11 (6.8) | 3 (10.7) | 8 (6) | — |
| inv(16)(p13.1q22)/t(16;16)(p13.1;q22) | 21 (13) | 16 (57.1) | 5 (3.8) | — |
AML, acute myeloid leukemia; AML-MRC, acute myeloid leukemia with myelodysplasia-related changes; ELN-2017, 2017 European Leukemia Net.
Baseline Characteristics of All Patients With De Novo AML and De Novo AML With or Without MRC-Related Mutations (Mutations in ASXL1, TP53, U2AF1, SRSF2 and SETBP1)
| Characteristic | De Novo AML (n = 161) | De Novo AML With MRC-Related Mutations (n = 28) | De Novo AML Without MRC-Related Mutations (n = 133) | P Value |
|---|---|---|---|---|
| Sex, male, No. (%) | 99 (60.7) | 20 (71.4) | 79 (59.4) | .23 |
| Age, median (range), y | 52 (14-88) | 60 (27-85) | 51 (14-88) | .001 |
| Bone marrow blasts, median (range), % | 58 (10-98) | 56 (13.5-94.5) | 58 (10-98) | .52 |
| WBCs, median (range), × 109/L | 9.4 (0.4-362.6) | 4.7 (0.36-94.16) | 11.6 (0.5-362.6) | .001 |
| Hemoglobin, median (range), g/L | 80 (42-159) | 79 (53-116) | 80 (42-159) | .43 |
| Platelets, median (range), × 1012/L | 39 (3-669) | 31 (5-196) | 42 (3-669) | .33 |
| ELN-2017 subgroup, No. (%) | ||||
| Favorable | 0 | 4 (14.3) | 43 (32.3) | — |
| Intermediate | 17 (10.6) | 11 (39.3) | 74 (55.6) | — |
| Adverse | 57 (35.4) | 13 (46.4) | 16 (12) | — |
| Not classified | 10 (6.2) | 0 | 0 (0) | — |
| Karyotype, No. (%) | ||||
| Normal karyotype | 85 (52.8) | 10 (35.7) | 75 (56.4) | — |
| t(8;21)(q22;q22) | 11 (6.8) | 3 (10.7) | 8 (6) | — |
| inv(16)(p13.1q22)/t(16;16)(p13.1;q22) | 21 (13) | 16 (57.1) | 5 (3.8) | — |
| Characteristic | De Novo AML (n = 161) | De Novo AML With MRC-Related Mutations (n = 28) | De Novo AML Without MRC-Related Mutations (n = 133) | P Value |
|---|---|---|---|---|
| Sex, male, No. (%) | 99 (60.7) | 20 (71.4) | 79 (59.4) | .23 |
| Age, median (range), y | 52 (14-88) | 60 (27-85) | 51 (14-88) | .001 |
| Bone marrow blasts, median (range), % | 58 (10-98) | 56 (13.5-94.5) | 58 (10-98) | .52 |
| WBCs, median (range), × 109/L | 9.4 (0.4-362.6) | 4.7 (0.36-94.16) | 11.6 (0.5-362.6) | .001 |
| Hemoglobin, median (range), g/L | 80 (42-159) | 79 (53-116) | 80 (42-159) | .43 |
| Platelets, median (range), × 1012/L | 39 (3-669) | 31 (5-196) | 42 (3-669) | .33 |
| ELN-2017 subgroup, No. (%) | ||||
| Favorable | 0 | 4 (14.3) | 43 (32.3) | — |
| Intermediate | 17 (10.6) | 11 (39.3) | 74 (55.6) | — |
| Adverse | 57 (35.4) | 13 (46.4) | 16 (12) | — |
| Not classified | 10 (6.2) | 0 | 0 (0) | — |
| Karyotype, No. (%) | ||||
| Normal karyotype | 85 (52.8) | 10 (35.7) | 75 (56.4) | — |
| t(8;21)(q22;q22) | 11 (6.8) | 3 (10.7) | 8 (6) | — |
| inv(16)(p13.1q22)/t(16;16)(p13.1;q22) | 21 (13) | 16 (57.1) | 5 (3.8) | — |
AML, acute myeloid leukemia; AML-MRC, acute myeloid leukemia with myelodysplasia-related changes; ELN-2017, 2017 European Leukemia Net.
For treatment response, the CR + CRi rate after the first round of standard-intensity induction therapy in AML-MRC was much lower than that in de novo AML (37.5% vs 77%, P < .001) Figure 4A. The CR + CRi rate for patients carrying AML-MRC–related mutations was close to that of patients with AML-MRC and much lower than that of patients without these mutations (38.5% vs 72.4%, P = .001) Figure 4B. In AML-MRC or de novo AML, respectively, the CR + CRi rates in patients carrying ASXL1, TP53, U2AF1, SETBP1, or SRSF2 mutations were inferior to those in patients without such mutations (in AML-MRC, 21.4% vs 50%; in de novo AML, 60% vs 78.9%) Figure 4C and Figure 4D.
A, Complete remission (CR) or CR with incomplete hematologic recovery (CRi) rates in acute myeloid leukemia with myelodysplasia-related changes (AML-MRC) vs de novo acute myeloid leukemia (AML) (P < .001). B, AML with (+) or without (–) MRC-related mutations (ASXL1, TP53, U2AF1, SRSF2, and SETBP1 mutations) (P =.001). C, AML-MRC with (+) or without (–) MRC-related mutations (P =.098). D, De novo AML with (+) or without (–) AML-MRC–related mutations (P =.34).
With regard to survival, carrying at least one AML-MRC–related mutation (ASXL1, TP53, U2AF1, SETBP1, or SRSF2 mutation) predicted unfavorable OS in all patients with AML (8.6 vs 29.1 months, P = .003) Figure 5A. Even in the ELN-2017 intermediate-risk group, which lacks the TP53 mutation and ASXL1 mutation without favorable changes, the presence of AML-MRC–related mutations still conferred shorter survival (10.6 months vs not reached, P = .03) Figure 5B. Patients with de novo AML with mutations in ASXL1, TP53, U2AF1, SETBP1, or SRSF2 had significantly inferior OS compared with other patients with de novo AML (15 months vs not reached, P = .003) Figure 5C. When comparing the outcomes of patients with de novo AML and those with AML-MRC with or without related mutations, the P value was not significant, and the Kaplan-Meier curves were similar Figure 5D (15 vs 8.2 vs 6.9 months, P = .49, P = .3). After adjusting for standard-intensity induction therapy, the presence of AML-MRC–related genes still indicated shorter survival with a trend toward significance (P = .075; HR, 1.77; 95% confidence interval, 0.95-3.32).
Survival outcomes. A, Overall survival (OS) curves for all patients with acute myeloid leukemia (AML) with (+) or without (–) myelodysplasia-related changes (MRC) (ASXL1, TP53, U2AF1, SETBP1, and SRSF2 mutations). B, OS curves for 2017 European Leukemia Net intermediate-risk patients with (+) or without (–) MRC-related mutations. C, OS curves for patients with de novo AML with (+) or without (–) MRC-related mutations. D, OS curves for patients with de novo AML and AML-MRC with (+) or without (–) MRC-related mutations. HR, hazard ratio.
Discussion
AML-MRC is a myeloid neoplasm that was first introduced as an acute leukemia subgroup in the 2008 WHO classification and further revised in the 2016 WHO classification.2 AML-MRC cases account for up to 48% of all adult-onset AML cases and generally feature old age, a low remission rate, and short OS.1 In this study, we compared the mutation landscapes between AML-MRC and de novo AML and identified distinct mutation patterns of these two groups. Specific mutations with high frequency in AML-MRC included ASXL1, NRAS, U2AF1, TP53, PTPN11, SRSF2, and SETBP1 mutations, while mutations in FLT3-ITD, WT1, KIT, NPM1, and CEBPA were typical in de novo AML. These two groups exhibited mutually exclusive mutation profiles. Among the AML-MRC–related mutations, mutations in chromatin structure (ASXL1) and splicing factor (U2AF1, SRSF2, and SETBP1) genes and TP53 confer shorter OS and AML-MRC–like features, even in patients with de novo AML. Our results indicate that NGS can aid in the assessment of patients with AML who exhibit AML-MRC–like features but do not meet any of the current AML-MRC diagnostic criteria.
Gene mutations in AML-MRC have been described in some studies.5,17,18 The mutations with high prevalence in our AML-MRC cohort—for example, those in genes encoding chromatin structure (ASXL1 and RUNX1), signal transduction (NRAS and PTPN11), RNA splicing (U2AF1, SRSF2, and SETBP1), DNA methylation (DNMT3A, IDH2, and IDH1), and cohesion (STAG2) and TP53—are consistent with previous reports.5,17 We confirmed that high-frequency mutations in AML-MRC were likely to co-occur, and some of these mutations were mutually exclusive with de novo AML mutations, suggesting that the two diseases are separate entities, which was also concluded in a large cohort study.19 Previous research found that mutations of WT1, KIT, and genes related to the RAS pathway, which are typical in de novo AML, usually appear in young and pediatric patients with AML,20 while AML-MRC–related mutations such as spliceosome mutations usually appear in older patients.19 This phenomenon is probably due to the high incidence of de novo AML in young patients but AML-MRC in older ones. These two mutation profiles may reveal two different disease incidence-related pathways that lead to distinct outcomes.
Alterations in DNMT3A, TET2, ASXL1, TP53, JAK2, and SF3B1 were commonly observed in clonal hematopoiesis of indeterminate potential (CHIP), and the mutation spectrum increased during the progression from CHIP to MDS and sAML.21 DNA methylation genes (DNMT3A, IDH2, and IDH1), chromatin structure genes (ASXL1 and RUNX1), RNA splicing genes (U2AF1, SRSF2, and SETBP1), cohesion genes (STAG2), and TP53 are usually mutated in MDS.22 For patients with sAML, some mutated genes, including TP53, RUNX1, STAG2, and ASXL1, have been proven to be acquired at the MDS stage and reserved when progressing to the sAML state.23 The similar mutation patterns between high-risk MDS and AML-MRC highlight the similar ontogeny between these myeloid malignancies overriding the diagnostic cutoff of 20% blast cells. Baer et al24 discovered that more than 10% of patients with AML without AML-MRC carried AML-MRC–like mutation patterns and had a dismal prognosis resembling that of patients with AML-MRC. Previous retrospective studies have confirmed that common mutations in AML-MRC, including ASXL1, TP53, and U2AF1 mutations, are related to poor survival.14 Therefore, we were able to identify a subset of patients with de novo AML who are similar to patients with AML-MRC in terms of inferior outcomes with the assistance of NGS.
Among the patients with AML-MRC, the rate of CR + CRi after the conventional “7 + 3” regimen is rather low at about 30%,3 which was further validated in our study. Of note, sAML-specific gene mutations (SRSF2, SF3B1, U2AF1, ZRSR2, ASXL1, EZH2, BCOR, or STAG2 mutations),7TET2, and DNMT3A mutations usually occur early in the phase before onset of AML and can be detected by NGS even during clonal remission.7,25 In our cohort, patients with ASXL1, TP53, U2AF1, SETBP1, or SRSF2 mutations, whether in AML-MRC or de novo AML, all had lower CR or CRi rates than those without such mutations. Resistance of blasts carrying particular mutations shared between MDS and AML-MRC to chemotherapies may partly explain the poor treatment response and survival in these patients. In view of their inherent resistance to chemotherapy, targeting and eliminating cancer cells carrying these mutated genes is a possible method for improving the survival of these patients.
There are some limitations of our study. First, it is a retrospective study, which inevitably means that the baseline information was not complete for all patients. Forty-two patients with AML in our cohort could not be classified due to missing karyotype results. Second, this is a single-center study with relatively few samples, which may have led to potential bias. Multicenter studies need to be conducted in a larger cohort to further verify our conclusions.
To conclude, we identified that patients with AML-MRC exhibit an entirely different mutation pattern from that of patients with de novo AML. Using AML-MRC–related genes can help us identify patients with AML who have inferior survival and lower remission rates at an early stage, supporting the development of a proper treatment strategy.
References
Author notes
First authors.
