Decitabine in combination with low-dose cytarabine, aclarubicin and G-CSF tends to improve prognosis in elderly patients with high-risk AML

Patient characteristics

The baseline characteristics of AML patients analyzed in this study are shown in Table 1. The study cohort consisted of 42 (33.6%) cases of AML with recurrent cytogenetic abnormalities (AML-RCA), 20 (16%) cases of AML with myelodysplasia-related changes (AML-MRC), nine (7.2%) cases of therapy-related AML (t-AML), and 54 (43.2%) cases of AML, not otherwise specified (AML, NOS). Cytogenetic examination revealed normal cytogenetics in 69 patients and complex karyotypes in 19 patients. The patients were classified into low-risk (12 patients; 9.6%), intermediate-risk (50 patients; 40.0%), and high-risk (63 patients; 50.4%) categories based on their cytogenetic characteristics and molecular abnormalities that were identified using NGS analysis of BM samples collected at diagnosis prior to any treatment.

Mutational analyses

NGS analysis revealed mutations in 19 different genes in 115 (92%) patients. Among these 115 patients, 34 (27.2%) had mutations in a single gene, 35 (28%) had mutations in two genes, 28 (22.4%) had mutations in three genes, 17 (13.6%) had mutations in four genes, and one had mutations in six genes. The most frequently mutated gene was TET2 (26.4%). The remaining mutated genes in the order of decreasing frequency were: ASXL1, 24.8%; FLT3, 23.2%; NPM1, 21.6%; DNMT3A, 19.2%; NRAS, 17.6%; CEBPA, 14.4%; IDH2, 12%; TP53, 12%; RUNX1, 8.8%; IDH1, 8.0%; and KIT, 5.6%; mutations in ETV6, SRSF2, U2AF1, PHF6, EZH2, CBL and CSF3R were rare (<5.0%); mutations were not identified in KRAS, JAK2, ZRSR2 and GATA2 genes (Figure 1A, Table 2). In comparison to patients with complex karyotypes, mutations in DNMT3A (21.7% vs. 0, P= 0.0341), TET2 (34.8% vs. 10.5%, P= 0.0485), NPM1 (14.5% vs. 0, P= 0.0047) and FLT3 (23.2% vs. 0, P= 0.0184) were more common in patients with normal karyotype. Conversely, the rate of TP53 mutations was significantly higher in patients with complex karyotypes than those with the normal karyotype (57.9% vs. 2.9%, P< 0.0001; Table 2). Ten out of fifteen patients (66.7%) with TP53 mutations belonged to the t-AML or the AML-MRC group. We divided the patients based on their age into three subgroups (60-66, 67-74 and 75-86 years) using ROC curve analysis. Mutations in IDH1 were more common in patients aged 67–74 years than those aged 60–66 years (18.0% vs. 3.1%, P=0.0248; Figure 1B). The rate of RUNX1 mutations were significantly higher in patients aged 75–86 years than those aged 60-66 year age group (14.3% vs. 4.6%, P= 0.0264; Figure 1B).

Clearance of mutations after induction treatment

We compared the frequency of mutations in the BM samples of 27 patients at diagnosis and after one cycle of induction treatment, and observed that clearance of leukemia-specific mutations correlated with morphological response and relapse. We observed significant reduction in variant allele frequency (VAF) of TP53, NPM1 and FLT3 mutations among responders after one cycle of D-CAG treatment, but no significant change was observed among the non-responders (Figure 2). Persistence of these gene mutations correlated with worse outcomes. Driver mutations persisted after chemotherapy and increased in size at the time of relapse. New mutations were also found at relapse, thereby suggesting that disease progression was coupled with step-wise genetic evolution.

Association between cytogenetics, gene mutations and clinical outcomes

After a median follow-up of 12 months (range: 2-82 months), 86 (68.8%) and 94 patients (75.2%) treated with D-CAG achieved CR after one cycle and two cycles of induction, respectively. Patients in the low- or intermediate-risk groups showed a higher CR rate than patients in the high-risk group, but the differences were not statistically significant (91.7% vs. 78.4% vs. 69.4%, P= 0.2053; Table 3).

At the final analysis on July 30, 2019, the median OS and DFS of all patients were 16 and 12 months, respectively. The median OS was 19, 14 and 9 months for patients aged 60–66 years, 67–74 years, and ≥75 years, respectively (P=0.007, Figure 3). The one-year and 2-year OS rates were 59.8% and 36.5%, respectively. The one-year and 2-year DFS rates were 49.3% and 28.2%, respectively. Patients aged 60–66 years showed significantly longer OS than patients ≥75 years (median OS: 19 months vs. 9 month, P= 0.0042, Table 3, Figure 3). However, DFS was statistically similar for both groups (median DFS: 15 months vs. 7 months, P= 0.0526, Table 3). The OS and DFS rates were significantly longer in patients belonging to the low- and intermediate-risk groups than in patients belonging to the high-risk group (median OS: Undefined vs. 20 months vs. 11 months, P= 0.0022; median DFS: 15 months vs. 15 months vs. 8 months, P= 0.0041, respectively; Table 3, Figure 4). Patients with complex karyotypes showed significantly shorter median OS and DFS compared to patients with non-complex karyotypes (n=19; median OS: 9 months vs. 19 months, P= 0.0041; median DFS: 5 months vs. 15 months, P= 0.0001, respectively; Table 3, Figure 5). Moreover, patients with abnormalities in chromosomes 5 and/or 7 (-5/5q- and/or -7/7q-) showed significantly shorter median OS and DFS compared to patients without these abnormalities (-5/5q- and/or -7/7q-, n=18; median OS: 8.5 months vs. 19 months, P< 0.0001; median DFS: 6 months vs. 15 months, P< 0.0001, respectively; Table 3, Figure 5). Patients with monosomal karyotypes (n=12) showed similar median OS, but significantly shorter median DFS compared to other patients (median OS: 13 months vs. 17 months, P= 0.3143; median DFS: 5 months vs. 14 months, P= 0.0004; Table 3, Figure 5).

Patients in the 60-66 and 67-74 age groups showed better OS and DFS for the low- and intermediate-risk group patients compared to the high-risk group patients. But, these differences were not statistically significant except that median DFS for the low- and intermediate-risk groups was significantly longer than the high-risk group for patients aged 67-74 years (median DFS: 19 months vs. 6 months, P= 0.0014; Figure 6).

We also analyzed 21 patients who were ≥75 years, including 8 low- or intermediate-risk patients and 13 high-risk patients. The median OS and DFS were relatively longer in the low- and intermediate-risk group compared to the high-risk group, but the differences were not statistically significant (median OS: 18 months vs. 9 months, P= 0.2110; median DFS: 13 months vs. 6 months, P= 0.1048; Figure 6). This suggests that D-CAG is feasible for the treatment of AML patients above 75 years, especially those harboring high-risk karyotypes and genetic mutations.

The median OS for patients with 0 or 1, 2 or ≥3 gene mutations was not statistically significant (12 months vs. 18 months vs. 17 months, P= 0.1570). Moreover, OS and DFS for patients with or without mutations in genes such as ASXL1 and RUNX1 were similar (Table 3). Patients with wild-type TP53 (n=110) showed relatively longer OS and DFS compared to patients harboring TP53 mutations (n=15), but the differences were not statistically significant (median OS: 18 months vs. 10 months, P= 0.0657; median DFS: 14 months vs. 7 months, P= 0.0649; Table 3, Figure 7). Patients with TP53 mutations (VAF <20%) showed relatively longer survival than patients with TP53 mutations (VAF ≥20%), but, the data was not statistically significant after excluding two patients with low VAF (<20%) TP53 mutations (median OS: 18 months vs. 9 months, P= 0.1047; median DFS: 14 months vs. 7 months, P= 0.0511). Ten out of 15 patients with TP53 mutations were associated with -5/5q- and/or -7/7q- chromosomal deletions, whereas, the remaining 5 patients had isolated TP53 mutations. The patients with isolated TP53 mutations showed relatively longer median OS and DFS compared to patients with TP53 mutations and concomitant -5/5q- and/or -7/7q- chromosomal deletions (median OS: undefined vs. 9 months, P= 0.0740; median DFS: 15 months vs. 7 months, P= 0.3662; Table 3, Figure 7). The median OS and DFS was comparatively similar for patients with wild-type and isolated TP53 mutations (median OS: 18 months vs. undefined, P= 0.5146; median DFS: 14 months vs. 15 months, P= 0.7177; Table 3, Figure 7), but, significantly lower in patients with TP53 mutations and concomitant -5/5q- and/or -7/7q- chromosomal deletions (median OS: 9 months vs. 18 months, P= 0.0016; median DFS: 7 months vs. 14 months, P= 0.0023, respectively; Table 3, Figure 7). The median OS and DFS was statistically similar for AML patients with or without FLT3 mutations (median OS: 12 months vs. 17 months, P= 0.2247; median DFS: 18 months vs. 12 months, P= 0.8653, respectively; Table 3, Figure 8). Patients with FLT3-ITD mutations in the absence of NPM1 mutations (n=8) and the other patients (n=117) showed statistically similar median OS and DFS (median OS: 12 months vs. 16 months, P= 0.3967; median DFS: 25 months vs. 12 months, P= 0.8556, Figure 8).

Univariable analyses showed that age (≥75 years), complex karyotypes, -5/5q- and/or -7/7q- chromosomal deletions, and high-risk status were independent prognostic factors associated with shorter OS (Table 4). Factors such as monosomal karyotypes, total number of gene mutations, mutated TP53 and FLT3, and FLT3-ITD mutations in the absence of NPM1 mutations did not show prognostic significance in the univariable analysis. Multivariate analysis showed that age (over 75 years), high-risk status, and -5/5q- and/or -7/7q- chromosomal deletions were significant variables that predicted poor prognosis or decreased OS (Table 4).

Adverse events

The most frequently observed adverse events during D-CAG therapy were myelosuppression and infections. All the AML patients treated with D-CAG experienced thrombocytopenia and neutropenia. Febrile neutropenia occurred in 82.4% of the patients, but sepsis was reported only in 6 cases. Non-hematological toxicities were usually mild to moderate (Table 5). For patients achieving CR, the median times for showing stable neutrophil (0.5×10⁹ neutrophils/L) and platelet (20×10⁹/L) counts were 14 days and 16 days, respectively.

Study populations

We enrolled 125 patients between the ages 60-86 years that were diagnosed with AML (except acute promyelocytic leukemia) in our hospital between September 2011 and October 2018 according to the WHO classification. The Eastern Cooperative Oncology Group Performance Status (ECOG PS) score of all patients was 0-3. The exclusion criteria included poor hepatic or renal function, defined as total serum bilirubin, aminotransferase, or creatinine concentrations more than two times the upper limit of normal range; poor cardiac function greater than Class II according to the New York Heart Association Functional Classification; presence of another malignancy without remission. The included patients had not received any previous chemotherapy other than hydroxyurea. The study procedures and informed consent forms were approved by the ethics committee of the First Affiliated Hospital of Nanjing Medical University, Jiangsu Province Hospital (number 2011-SR-085). We obtained informed consent from all patients included in this study or their legal guardians.

This study was registered on the Chinese Clinical Trial Registry (ChiCTR No. 11001700). The target accrual number for this study was 100 and the duration was from September 2011 to September 2016. We enrolled 89 patients in this observational study (ChiCTR number 1001700), including 81 patients that participated in our previous study [8]. The remaining 36 patients came from an extension cohort, including 16 patients aged above 75 years. They were not enrolled in the clinical trial and were treated with D-CAG according to the 2017 Chinese guidelines for diagnosis and treatment of adult acute myeloid leukemia [35]. These guidelines required AML patients above 60 years to receive decitabine combined with low-dose chemotherapeutic regimen, such as CAG.

D-CAG treatment regimen

The patients were treated with the D-CAG regimen as previously reported [8]. Briefly, patients were intravenously injected with 15 mg/m² decitabine over 4 h for five consecutive days (days 1-5), 10 mg/m² cytarabine every 12h for seven days (days 3-9), 10 mg/m²/day aclarubicin for four days (days 3-6) from October 2010 to April 2016 or 8 mg/m²/day aclarubicin for four days (days 3-6) from May 2016 to May 2018, and 300 μg/day G-CSF for priming until the white blood cell (WBC) counts exceeded 20×10⁹/L. Up to two cycles of induction chemotherapy were allowed if CR was not achieved. Patients who did not achieve CR after two cycles of D-CAG were offered alternative therapies or supportive therapy. The post-remission therapy included 4–6 cycles of D-CAG for the patients achieving CR. Upon final analysis on July 30, 2019, D-CAG as post-remission therapy was administrated in 89 patients for one cycle, 74 patients for two cycles, 40 patients for three cycles, 33 patients for four cycles, 12 patients for 5 cycles and 5 patients for 6 cycles, respectively.

Cytogenetic analyses

Bone marrow (BM) cells were directly harvested from unstimulated cultures. Metaphase cells were banded using an improved heat treatment and Giemsa R-banding method. The karyotyping was based on conventional cytogenetic examination of ≥20 metaphases.

Next-generation sequencing

A total of 23 genes, including DNMT3A, TET2, IDH1, IDH2, SRSF2, U2AF1, ZRSR2, ASXL1, EZH2, RUNX1, NPM1, CEBPA, ETV6, GATA2, FLT3, NRAS, KRAS, CBL, JAK2, KIT, CSF3R, TP53 and PHF6, were included in the targeted gene sequencing (TGS) panel. Genomic DNA (gDNA) was extracted from the BM aspirates of each patient using an Autopure extractor (Qiagen, Hilden, Germany). Then, 10 ng genomic DNA was PCR amplified using the Ion AmpliSeq Library kit 2.0 (Ion Torrent, Thermo Fisher Scientific, USA). The 206 amplicons included in this panel are summarized in Supplementary Table 1. Amplicon libraries were constructed using the KAPA Hyper Prep kit for Illumina Platforms (Kapa Biosystems, Wilmington, MA, USA) [36], and sequenced on the Illumina Miseq platform. The sequencing reads were aligned to the human reference genome (genome build hg19) using the Burrows-Wheeler Aligner (BWA) [37]. Variant calling was performed using the Genome Analysis Toolkit (GATK) [38]. The Integrative Genomics Viewer (IGV) version 2.3.32 was used to visualize the sequencing reads and assess the variants [39]. We excluded from further analysis the synonymous and non-synonymous variants that occurred with a frequency >0.1% in the 1000 Genomes Project database (http://www.ncbi.nlm.nih.gov/variation/tools/1000genomes/) or in the normal Eastern Asian population from the Exome Aggregation Consortium (ExAC; http://exac.broadinstitute.org/).

Definition of outcomes

Treatment responses were assessed according to the NCCN clinical practice guidelines of AML (version 2.2019). The complete remission (CR) was defined by transfusion independence, <5% blasts in the BM aspirate with spicules, absence of blasts with Auer rods, absence of residual extramedullary disease, as well as absolute neutrophil counts >1.0×10⁹/L, and platelet counts ≥100×10⁹/L in peripheral blood. Partial remission (PR) was defined by a 50% decrease in the percentage of blasts to 5-25% in the BM aspirate, and the normalization of blood counts as described above. No remission (NR) was indicated when the CR or PR criteria were not met as described above. The objective response rate (ORR) included rates of CR and PR. The overall survival (OS) was measured from the time of diagnosis to death or censorship at the last follow-up. The disease-free survival (DFS) was defined as the duration from CR until relapse or death or censorship at the last follow-up. The time to stable neutrophil recovery was measured from the end of protocol induction therapy until the first day when the absolute neutrophil counts recovered to ≥500/μl for two consecutive measurements on different days. The time to stable platelet recovery was measured from the end of protocol induction therapy until the first day that the platelet counts were ≥20,000/μl for at least seven consecutive days. Toxicities were defined and graded according to the National Cancer Institute (NCI) Common Toxicity Criteria [40].

Statistical analysis

Data were analyzed using the Statistical Package for Social Sciences (SPSS version 20.0). Statistical significance was considered when the p value was <0.05. Differences in continuous variables were analyzed by t-tests and chi-squared tests. Fisher exact test was performed to compare the incidences. Kaplan–Meier analysis was performed to estimate the survival probabilities, and proportional hazards model was used for univariate and multivariate analysis.