Introduction

Genetic mutation is a frequent molecular event in acute myeloid leukemia (AML) and various mutations increase cell proliferation, block differentiation, and determine prognosis [1]. FMS-related tyrosine kinase 3 (FLT3) is a member of type-III tyrosine kinase receptor that plays an important role in proliferation and differentiation of early hematopoietic progenitor cells [2]. Internal tandem duplications of the juxtamembrane domain of FLT3 (FLT3-ITD) are among the most common mutations in AML, about 30% [3]. Resulting in constitutive activation of the kinase, AML patients with FLT3-ITD mutation often have a high peripheral leucocyte count and bone marrow blast cell count, and increased rates of relapse, implying that FLT3-ITD mutation portends a particularly poor prognosis of AML [4, 5]. Moreover, DNMT3A mutations, especially DNMT3A R882H mutation leading to reduced methyltransferase activity, were often found in AML with normal karyotype, particularly in AML subtypes M4 and M5, with a mutation proportion of about 20% [6, 7]. AML patients with DNMT3A mutation usually present a poor prognosis [8]. Large-scale genomic sequencing studies have identified a subset of patients featured with concomitant FLT3-ITD and DNMT3A mutations [9, 10]. Furthermore, the co-occurrence of these two mutations is significantly associated with poor clinical outcomes by a low first complete remission (CR) rate and a quick relapse after first CR [10]. However, the prognosis of these AML patients with the double mutation after allogeneic hematopoietic stem cell transplantation (allo-HSCT) have been less studied. The present study retrospectively analyzed the clinical outcomes of a subgroup of Chinese AML patients with this double mutation after chemotherapy and allo-HSCT treatment.

Materials and methods

Patients

We retrospectively studied 206 patients with AML who underwent targeted Next-generation sequencing (NGS) of leukemia samples and received related or unrelated allo-HSCT from January 2010 and December 2014 at the First Affiliated Hospital University of Soochow University. But M3, >65 years old, patients who had not received CR before transplantation and patients with FLT3-ITD-positive AML who received an FLT3 inhibitor before and/or after transplantation were excluded. There were 107 men and 99 women with a mean age of (37.3 ± 11) years. Including 6 cases of M0, 24 cases of M1, 56 cases of M2, 39 cases of M4, 63 cases of M5, 6 cases of M6, and 12 unclassified cases. The baseline of patient characteristics are listed in Table 1. The institutional review board (IRB) of our hospital approved the study, and all patients or their guardians signed consent forms approved by the IRB.

Table 1 Comparison of clinical data between four groups (number of cases)
Full size table

Chemotherapy

After diagnosis, all patients received 3 + 7 standard induction chemotherapy: idarubicin 8–12 mg/m2 for Day 1–3 and cytarabine 100 mg/m2 for Day 1–7. After achieving CR by induction chemotherapy, patients were given 1–2 courses of consolidation chemotherapy based on FA (Fludarabine 30 mg/m2qd d1-5 and Ara-c1-2 g/m2qd d1-5) or an intermediate dosage of Ara-C1-2 g/m2 q12 h d1-3. Subsequently, patients underwent allo-HSCT if they had a suitable donor. In addition, refractory/relapsed patients were given salvage chemotherapy, such as FLAG (Fludarabine30 mg/m2qd d1-5, Ara-c1-2 g/m2qd d1-5, and G-CSF300ug/m2d0-5) and MAE (Mitoxantrone 10 mg/m2qd d1-3, Ara-c100-200 mg qd d1-7, and Etoposide 100 mg qd d1-3), and they underwent allo-HSCT as soon as possible if there was an appropriate donor once they achieved CR.

allo-HSCT

Indication for allo-HSCT

All patients were divided into three groups as follows: low, intermediate, and poor groups according to their biological and genetic characteristics [11]. Patients in the intermediate and poor groups received allo-HSCT at their first CR. Patients in the low-risk group received allo-HSCT only in the situation of minimal residual disease (MRD) persistently increased. All patients in status more than CR2 received allo-HSCT.

Transplantation procedure

In addition to 16 cases with important organ dysfunction received reduced-intensity conditioning regimens [12]. The rest received myeloablative conditioning [13, 14]. All transplantations used T-cell–replete grafts and standard calcineurin inhibitor-based graft-versus-host disease prophylaxis [12]. Post-transplantation maintenance was not routinely used.

Gene mutation analyses

Patients’ bone marrow specimens at diagnosis were retained in the specimen bank of the hospital. Genomic DNA was extracted using genomic DNA Purification Kit (Progma) following the manufacturer’s protocol. The exons 8 and 17 of C-Kit gene, exon 12 of NPM1 gene, exon 20 of FLT3 gene, 882 mutation hotspot region of DNMT3A gene, all exons of CEBPA gene were amplified by PCR using Hot Start Taq enzyme (ABI), where PCR reaction conditions were: 95 °C for 5 min; 95 °C for 30 s, 58 °C for 30 s, 72 °C for 90 s, 35 cycles; 72 °C for 7 min, and stored at 4 °C. The primers are shown in Supplemental Table X. PCR products were sequenced for mutation detection with an ABI-3730 genetic analyzer.

Follow-up

Follow-up was processed through their regular visit to the outpatient department and questionnaire or telephone interview. The main endpoints of this study were leukemia-free survival (LFS) and overall survival (OS), where LFS refers to the duration from post-transplantation to recurrence of leukemia or death, and OS refers to the time from post-transplantation to death for any reason.

Statistical analysis

Clinical data were analyzed among the four groups using Chi-square test and analysis of variance. OS and LFS were analyzed using the Kaplan–Meier method. The incidences of relapse, NRM, aGVHD, and cGVHD were calculated using the cumulative incidence method, considering competing risks by R software. Univariate analyses of relevant prognostic factors after allo-HSCT were performed using the log-rank test. Multivariate analyses were performed using the COX regression model. A difference with p < 0.05 was considered statistically significant.

Results

Mutant genes detected in all AML patients at diagnosis

Mutant genes detected in 206 AML patients are shown in Table 2, from which the top five genes sorted by mutation ratio were: FLT3-ITD + (12.6%), double mutation CEBPA(duCEBPA) (9.7%), DNMT3A R882 + FLT3-ITD + NPM1 + (6.8%), DNMT3A R882 + (6.8%), single mutation CEBPA(siCEBPA) + (4.4%), and FLT3-ITD + NPM1 (3.4%). While, the percentage of mutation negative was 39.8%.

Table 2 Mutant genes in 241 AML patients receiving allo-HSCT detected at diagnosis
Full size table

Grouping according to genetic mutation

According to the results of genetic mutation, 206 patients undergoing allo-HSCT were divided into four groups: Group A (FLT3-ITD + DNMT3A R882+) consisted of 19 AML patients with FLT3-ITD + DNMT3A R882 ± other mutations (CEBPA or NPM1); Group B (FLT3-ITD + DNMT3A R882−) consisted of 38 patients with FLT3-ITD ± other mutations (CEBPA or NPM1 but without DNMT3A mutation); Group C (FLT3-ITD- DNMT3A R882+) consisted of 21 patients with DNMT3A R882 ± other mutations (CEBPA or NPM1, but without FLT3-ITD mutation); and Group D (FLT3-ITD- DNMT3A R882- group) consisted of 128 patients with neither FLT3-ITD mutation nor DNMT3A R882 mutation (might be accompanied by other mutations such as CEBPA, NPM1, C-kit, and FLT-TKD). The 206 patients were followed-up for a median duration of 25 (1–58) months. Until the endpoint of follow-up, in group A, there were 12 recurrent cases (63.2%) and 12 death cases (63.2%). Of the death cases, 5 were (26.3%) from recurrence, 4 (21.1%) from severe GVHD, 2 (10.5%) from infection, and 1 case (5.3%) from suicide. More detail are shown in Table 3.

Table 3 Clinical outcomes in four groups (number of cases)
Full size table

Clinical features of AML with FLT3-ITD + DNMT3A R882+ double mutation

Of the 206 AML patients, 19 (7.8%) were double mutation, including 8 cases of M5 (42.1%), 5 cases of M2 (26.3%), 4 cases of M4 (21.1%), and 2 cases of unclassified AML (10.5%). These patients had a median white blood cell count of 38 (10.8–279) × 109/L. The ratio of hyperleukocytosis (≥50 × 109/L) of Group A (42.1%) at diagnosis was significantly higher than that of the group D (17.7%) (p = 0.014). 18 (94.7%) of the 19 AML patients with double mutation had a normal chromosomal karyotype, and 1 (5.3%) had a karyotype of 47,XY, + 21 [3] /46,XY [7]. 4 (21.1%) of the 19 AML patients were only FLT3-ITD + DNMT3A R882 + double mutation. 14 of 19 (73.7%) were FLT3-ITD + DNMT3A R882 + NPM1 + and 1 of 19 (5.3%) was FLT3-ITD + DNMT3A R882 + NPM1 + c-kit+. 10 (52.6%) of 19 AML patients with double mutation achieved CR after the first induction chemotherapy, which was significantly lower than that of the group D (77.7%) (p = 0.019). However, regarding CR rate, there were no difference between group A, group B, and group C (52.6 vs 71.1 vs 76.2%, p > 0.05). 14 (73.7%) patients received allo-HSCT at CR1, 3 (15.8%) at CR2, and 2 (10.5%) at ≥CR2. With a median follow-up time of 9 months (2–36) after transplantation, the cumulative incidences of aGVHD and cGVHD in group A were 36.8 and 27.6%, respectively. 7 patients relapsed and 12 died; the cumulative incidence of transplant-related mortality (TRM) of group A (16.5%) was significantly higher than that of group D (4.2%) (p = 0.016). But there was no difference between group B (13.6%) and group C (6.4%) (p = 0.911 and p = 0.575, respectively). Of the 12 death, 5 (26.3%) died from recurrence within 6 months, 4 (21.1%) from severe GVHD, 2 (10.5%) from infection, and 1 (5.3%) from suicide. The 6-month CIR (36.8%) was significantly higher than that of group D (6.1%) (p < 0.001), and higher than those of group B (26.3%) and group C (15.4%), but the difference was not significant between group A, group B, and group C (p = 0.340 and p = 0.094, respectively).

Comparison of clinical characteristics among different groups

Except for gender, leukocyte count at diagnosis and chromosome karyotype, the median age, FAB classification, disease status prior to transplantation, type of donor, conditioning regimen, and GVHD were not significantly different between four groups (Table 1). The cytogenetics analyses of all patients were normal with the exception of one patient with +21 in group A. There were 35 cases of intermediate-risk karyotypes (27 cases of normal karyotypes and 8 cases of abnormal karyotypes) and 3 cases of poor karyotypes in group B. All patients showed a normal karyotype in group C with the exception of one case with +21. There were 105 cases of intermediate-risk karyotypes and 23 cases of poor karyotypes in group D (Table 4). Risk stratification of chromosome refers to NCCN Guideline (version 2012) [11].

Table 4 Chromosome karyotype and risk stratification in four groups
Full size table

Comparison of 2-year CIR, OS, and LFS among four groups

The 2-year CIR rate of the group A (72.2%) was significantly higher than those of the group B (38.6%) and group C (36.8%), while the 2-year OS rate and LFS rate (30.9 and 11.3%, respectively) were significantly lower than those of the group B and group C (2y-OS 67.5%, 61.4%; 2y-LFS 47.9%, 56.8%, respectively). Meanwhile, the 2-year CIR rates of the group B and group C were significantly higher than that of the group D (27.8%), while their OS and LFS rates were significantly lower than that of the group D (2y-OS 80.5%; 2y-LFS 80.6%). The 2-year CIR rate, OS rate, and LFS rate of the group B and C had no statistical difference as shown in Figs. 1, 2, 3.

Fig. 1
figure 1

R software analysis of CIR rate among four groups

Full size image
Fig. 2
figure 2

Kaplan–Meier analysis of OS rate between the four groups

Full size image
Fig. 3
figure 3

Kaplan–Meier analysis of LFS rate between the four groups. The 2-year CIR rate of the group A (72.2%) was significantly higher than other groups (group B 38.6%, group C 36.8%, group D 27.8%), p < 0.05. While the 2-year OS rate and LFS rate of the group A (30.9 and 11.3%, respectively) were significantly lower than other groups (group B 67.5%, 47.9%, group C61.4%, 56.8%, group D 82.5%, 80.9%), p < 0.05

Full size image

Univariate and multivariate analyses of the prognostic factors

Univariate analyses were performed for prognostic factors (including gender, age, leukocyte count at diagnosis, FAB classification, chromosome, disease status prior to transplantation, type of donor, conditioning regimen, and gene mutation) associated with OS of AML patients undergoing allo-HSCT and the results revealed that the age, leukocyte count at diagnosis, disease status prior to transplantation, FLT3-ITD, and FLT3-ITD + DNMT3A R882+ were associated with poor outcome after transplantation. These factors with p < 0.100 were then analyzed by multivariate COX regression analyses with OS, and the results indicated that the disease status prior to transplantation, FLT3-ITD, as well as FLT3-ITD and DNMT3A R882 double mutation were independent factors of poor prognosis after transplantation (Tables 5, 6).

Table 5 Univariate analyses of prognostic factors and OS in 241 AML patients receiving allo-HSCT
Full size table
Table 6 Multivariate COX regression analyses of prognostic factors in AML patients receiving allo-HSCT
Full size table

Discussion

Although the prognostic impact of molecular abnormalities are relatively well understood in the setting of chemotherapy-treated AML patients, less is known about their role in determining prognosis after allo-HSCT. FLT3-ITD and DNMT3A R882 mutations are both factors for poor prognosis of AML. Allo-HSCT has been confirmed to improve the prognosis of AML with either of gene mutations [15, 16]. However, the prognosis of AML patients with FLT3-ITD and DNMT3A R882 double mutations especially in the population of chinese after transplantation has been rarely reported. To investigate the clinical characteristics of these AML patients with double mutations and their outcome of transplantation, we examined 6 frequently mutated genes (including FLT3-ITD, DNMT3A, C-kit, CEBPA, FLT3-TKD, and NPM1) in 206 pre-transplantation samples from patients with AML using direct sequencing method in our center. The results showed that comparing with other three groups of AML patients, the outcome of FLT3-ITD + DNMT3A R882+ group was the worst in terms of 2-year OS, 2-year LFS as well as CIR. And being in remission at the time of transplantation did not eliminate the increased relapse risk.

As we known, FLT3 is a member of the type-III tyrosine kinase receptor family, and plays an important role in proliferation and differentiation of early hematopoietic progenitor cells. FLT3 has two meaningful mutation patterns: ITD and TKD [17]. AML patients with FLT3-ITD mutation had been found to have a poor prognosis [18]. Although, the CR rate of AML patients with FLT3-ITD mutation after induction chemotherapy was similar to that of AML patients without FLT3-ITD mutation, the AML patients with FLT3-ITD mutation had a high risk of early recurrence even if they received a high dose of Ara-C consolidation chemotherapy after CR [19, 20]. The mechanism for the recurrence was still unclear. Therefore, it has been widely recognized that patients with FLT3-ITD mutation achieving CR1 should undergo allo-HSCT as soon as possible [5]. However, many previous studies found that allo-HSCT does not override the poor prognostic feature of FLT3-ITD positive with a higher risk of relapse and worse survival. Schmid C et al. analyzed the prognosis of 702 adult AML patients with normal chromosomes who received allo-HSCT after the first CR, including 75 patients with FLT3-ITDmut /NMP1 wt and 269 patients with FLT3-ITDwt /NMP1 mut, and they found that the 2-year OS rates of these two groups were (54 ± 7) % and (66 ± 3) %, respectively. Furthermore, multivariate analysis revealed that FLT3-ITD was one of the poor prognostic factors [21]. Luskin MR et al. retrospectively studied pre-transplantation genetic profiles obtained from next-generation sequencing of 26 genes in 112 adult patients with AML who underwent allo-HSCT. They found that mutations in TP53, WT1, and FLT3-ITD were associated with an increased risk of relapse after allo-HSCT [22]. Jae-Sook Ahn’s group also confirmed that the FLT3-ITD pos group had worse survival after allo-HSCT than those without FLT3-ITD [23]. Our study also revealed that FLT3-ITD was an independent factor for poor prognosis of AML after allo-HSCT.

Moreover, DNMT3A mutant AML patients with normal cytogenetics often had a poor prognosis [24]. About 50% DNMT3A mutations happened at codon R882 [25]. AML patients with DNMT3A R882 mutation had an adverse prognosis compared to AML patients with DNMT3A non-R882 mutation [26]. DNMT3A R882 mutation suggested a lower recurrence-free survival and a lower overall survival in AML patients [27]. Jae-Sook Ahn et al. compared prognoses of DNMT3A R882, non-R882 DNMT3A, and wild-type DNMT3A groups after allo-HSCT, and found that the 5-year overall survival and event-free survival of the DNMT3A R882 group were significantly decreased while 5-year recurrence was significantly elevated compared with the other two groups [23]. Our results also confirmed that although FLT3-ITD-DNMT3A R882+ group had a better 2-year OS and 2-year LFS than double mutation group, but it was lower than FLT3-ITD-DNMT3A R882 group. However, the FLT3-ITD- DNMT3A R882+ groups had a relative higher 2-year CIR than that of the FLT3-ITD-DNMT3A R882- group. So far, the mechanism of DNMT3A R882 mutation associated with the poor outcome of AML was still unclear. But it is speculated that DNMT3A R882 mutation might suppress DNA methyltransferase more significantly compared with the DNMT3A non-R882 mutation [28]. Recently, research of Yang L et al. showed that DNMT3A loss drives enhancer hypomethylation in FLT3-ITD-associated leukemias [29].

Double mutation with FLT3-ITD and DNMT3A mutation concomitant had been frequently found in AML patients [30]. FLT3-ITD mutation was seen in 41.7% of DNMT3A mutant AML patients compared with 23.7% in AML patients without DNMT3A mutation. Furthermore, AML patients with FLT3-ITD and DNMT3A double mutation had a low CR rate, high risk of recurrence after remission after induction chemotherapy as well as allo-HSCT [9]. Our findings revealed that the 2-year OS and LFS of AML patients with FLT3-ITD and DNMT3A R882 double mutation after allo-HSCT were significantly lower than that of single mutation groups with FLT3-ITD or DNMT3A. But double mutation group had a higher cumulative incidence of relapse rate than that of single mutation groups. Furthermore, multivariate analyses revealed that FLT3-ITD + DNMT3A R882+ double mutation was an independent factor for poor outcome post-transplantation. This finding was similar with a previous study reported by Jae-Sook Ahn et al. In their study on 115 AML patients receiving allo-HSCT after CR, where the DNMT3A R882 and FLT3-ITD double mutation was an independent factor of poor prognosis in 5-year overall survival (9.1%) and event-free survival (9.1%) [23]. In recent years, chemotherapy with Decitabine and Sorafenib had improved the outcomes of AML patients with a single mutation. Further investigation will be studied whether these DNMT3A R882 and FLT3-ITD double mutant AML patients will benefit from these target therapy or not.

Undoubtedly, the main limitations of our study are its small sample size, heterogeneity of the cohort, and retrospective nature; therefore, these results require prospective validation in larger cohorts.

In summary, Chinese AML patients with FLT3-ITD and DNMT3A R882 double mutation had a poor prognosis even after allo-HSCT manifesting as low OS and LFS as well as a high CIR rate. Therefore, it is necessary to explore prevention strategy for recurrence after transplantation.