DNMT3A mutations in Chinese childhood acute myeloid leukemia

Abstract Background: DNA methyltransferase 3A (DNMT3A) mutations have been found in approximately 20% of adult acute myeloid leukemia (AML) patients and in 0% to 1.4% of children with AML, and the hotspots of mutations are mainly located in the catalytic methyltransferase domain, hereinto, mutation R882 accounts for 60%. Although the negative effect of DNMT3AR882 on treatment outcome is well known, the prognostic significance of other DNMT3A mutations in AML is still unclear. Here, we tried to determine the incidence and prognostic significance of DNMT3A mutations in a large cohort in Chinese childhood AML. Methods: We detected the mutations in DNMT3A exon 23 by polymerase chain reaction and direct sequencing in 342 children with AML (0–16 years old) from January 2005 to June 2013, treated on BCH-2003 AML protocol. The correlation of DNMT3A mutations with clinical characteristics, fusion genes, other molecular anomalies (FLT3 internal tandem duplication [FLT3-ITD], Nucleophosmin 1, C-KIT (KIT proto-oncogene receptor tyrosine kinase), and Wilms tumor 1 mutations), and treatment outcome were analyzed. Results: DNMT3A mutations were detected in 4 out of 342 (1.2%) patients. Two patients were PML-RARA positive and 1 patient was FLT3-ITD positive. The mutations in coding sequences included S892S, V912A, R885G, and Q886R. Furthermore, there was 1 intronic mutation (c.2739+55A>C) found in 1 patient. No association of DNMT3A mutations with common clinical features was found. Two patients with DNMT3A mutations died of relapse or complications during treatment. One patient gave up treatment due to remission induction failure in day 33. Only 1 patient achieved continuous complete remission. Conclusions: DNMT3A mutations were rare in Chinese children with AML including PML-RARA positive APL. The mutation positions were different from the hotspots reported in adult AML. DNMT3A mutations may have adverse impact on prognosis of children with AML.


Patients
A total of 342 children with AML, including 3 cases with acute myeloblastic leukemia, without maturation (M1), 142 cases with acute myeloblastic leukemia with granulocytic maturation (M2), Editor: Ahmet Emre Eskazan. ZL designed the study; WL, CG, LC, SL, and XZ collected samples and analyzed some data; MW and RZ provided the clinical data and therapy program.
This work was supported in part by a grant-in-aid from a project of The Tutoring Project for National Natural Science Foundation in Beijing Children's Hospital (GQN201614); National Natural Science Foundation of China (No. 81170504, 81200392), Beijing Municipal Natural Science Foundation (No. 7152054, Study on the mechanism of relapse resulting from low expression of CASP8AP2 in acute B lymphoblastic leukemia), Beijing Health Qualified Personnel Program (No. 2011-3-049). 76 cases with acute promyelocytic leukemia (M3), 43 cases with acute myelomonocytic leukemia (M4) or together with bone marrow eosinophilia (M4Eo), 33 cases with acute monoblastic leukemia (M5), 11 cases with acute erythroid leukamia (M6), 29 cases with acute megakaryoblastic leukemia (M7), and 5 cases with mixed phenotype acute leukemia (MPAL), were enrolled in this study. The patients were diagnosed as AML from January 2005 to June 2013 at Beijing Children's Hospital. There were 208 boys and 134 girls, aged from 1 to 16 years with a median of 7 years. One hundred and ninety-seven patients were found positive for different types of fusion genes including AML1-ETO, PML-RARA, and CBFb-MYH11, MLL rearrangements, DEK-CAN and TLS-ERG fusion. The patients were treated in accordance with the AML BCH-2003 Protocol. Informed consents were obtained from all the children's parents or legal guardians.

Nucleic acid extraction
Ficoll 400 (MD Pacific Technology Co., Ltd., Tianjin, China) was used to isolate bone marrow mononucleated cells. The cells were stored at À70°C until use. We used the DNA Extraction Kit (Ugene Co., Ltd., Anhui, China) to extract genomic DNA. Trizol Reagent (Invitrogen, Carlsbad, CA, Promega, Madison, WI) was used to extract total RNA. Then the RNA was reverse transcribed into cDNA with moloney murine leukemia virus reverse transcriptase (Invitrogen; Promega). [22]

Detection of DNMT3A mutations
In this study, the mutations in exon 23 as well as adjacent intronic regions were focused, because most of DNMT3A mutations were concentrated in exon 23. The polymerase chain reaction (PCR) mixture was 50 mL, containing 100 ng of genomic DNA, 5 mL of 10 times buffer, 1.5 mM MgCl 2 , 0.2 mM dNTPs, 10 pmol of upstream and downstream primers, and 1 to 2 U of Taq DNA polymerase (Promega, USA). The primers' sequences could be found in Table 1. The cycling condition was as follows: 5 minutes at 95°C for predenaturation, 40 cycles of 30 seconds at 95°C, 30 seconds at 55°C, and 30 seconds at 72°C, and 10 minutes at 72°C for final extension. The PCR products were sent to Shanghai Sangon Biological Engineering Technology & Service Co., Ltd. and directly sequenced using AB PRISM 3730 Automated Sequencer. Four PCR products with unsatisfactory sequencing results were subcloned into pEASY TM -T5 Zero cloning vector. The recombinant plasmids were transformed into Trans 5a chemically competent cell (Transgene, Beijing, China). Five to 10 clones for each product were selected for plasmid sequencing after incubation at 37°C for 15 hours. [22] 2.4. Analysis and interpretation of DNMT3A mutations Mutation Surveyor software (SoftGenetics, PA) was used to analyze the sequencing results for existence of mutations. Variant effect predictor (Ensembl, Sanger Institute, Cambridge, United Kingdom) online tool and mutation taster online tool were used to analyze the possible effect on protein function for every mutation of DNMT3A. The National Center for Biotechnology Information record NM_175629 was used as the reference sequence of DNMT3A.

Statistical analysis
The difference in clinical characteristics between the 2 groups of patients with or without DNMT3A mutations was tested using Fisher exact test. SPSS 16.0 software (SPSS Inc., Chicago, IL) package was used to statistically analyze all of the data. P < .05 was appointed statistically significant.

DNMT3A mutations in childhood AML
In total, 4 out of 342 (1.2%) patients with newly diagnosed AML were identified with DNMT3A mutations by direct sequencing and clone sequencing for exon 23. PML-RARA fusion gene was detected in 2 patients' leukemic blasts. No other fusion gene was found in the other 2 patients. Three missense mutations and one synonymous mutation in exon 23 including V912A, R885G, Q886R, and S892S were detected in 4 patients respectively ( Fig. 1 and Table 2). In patient 3, we also detected one mutation in intronic regions: c.2739+55A>C (Table 2). Only S892S has been reported previously, [13] the other 4 were all novel variants (Table 3). However, the most common mutation R882 was not found in our series of cases. Of note, the DNMT3A mutations detected in diagnostic samples did not exist in corresponding complete remission (CR) samples. These findings indicated that these mutations were leukemia specific and were not germline. Table 1 Primers for mutation detection in childhood AML. In order to assess the possible effect of the 5 mutations on protein structure and function, we analyzed the mutations with variant effect predictor and mutation taster. The 3 missense mutations, V912A, R885G, and Q886R, changed amino acid sequence as well as splice site, with moderate impacts. In addition, the latter 2 might affect protein features. Although S892S and c.2739+55A>C made no effect on amino acid sequence, these 2 mutations changed splice site. Interestingly, the former might also affect protein features according to analysis result of mutation taster (Table 3).

Correlation of DNMT3A mutations with clinical characteristics
We firstly analyzed the association of DNMT3A mutations with other 4 gene (NPM1, C-KIT, WT1, and FLT3-ITD) mutations respectively, but found no significant correlation with them (Fisher exact test, P > .05, Table 4). Furthermore, no correlation was found between the common clinical features including age, sex, diagnostic white blood cell count, leukemia subtype, karyotype, fusion genes, and DNMT3A mutations (Table 4).

Clinical outcome of patients' with DNMT3A mutations
DNMT3A mutations were detected in 4 out of 342 (1.2%) patients. Of those with DNMT3A mutated, patient 1 (man, 6 years old, FAB M4) harboring a S892S mutation was diagnosed as leukemic infiltration of the gastrointestinal tract, and gave up treatment after 38 days due to remission induction failure at day 33. Patient 2 (woman, 12 years old, APL) harboring a V912A mutation and PML-RARA fusion gene achieved continuous complete remission for 60 months. Patient 3 (man, 1 year old, FAB M5) harboring a R885G mutation and 1 intronic mutations (c.2739+55A>C) was diagnosed as AML with testicular and CNS involvement. After 10 months, the patient undertook hematopoietic stem cell transplant but relapsed again in testicle after 39 days and died 8 months later. It was suggested that complex mutations of DNMT3A may associated with poor prognosis. It was noteworthy that 2 mutations (R885G and c.2739+55A>C) detected at diagnosis could not been found at testicular relapse in patient 3. Thus, it was possible that mutations in other genes might occur in the relapse clone. Patient 4 (woman, 2 years old, APL) harboring a Q886R mutation was found to be PML-RARA and FLT3-ITD positive, and died of retionic acid syndrome 22 days later ( Table 2).
S892S mutation changed the splice site and might affect the protein features according to mutation taster tool (University of Berlin,Germany). However, Pt.1 with DNMT3A [S892S] gave up treatment due to leukemic infiltration of the gastrointestinal tract in the early phase of treatment, we were not sure about the association of S892S with prognosis. V912A mutation was a fairly subtle amino acid change and did not affect the protein features ( Table 3). As codon 912 was at the end of the protein (stop codon was at 913), it is difficult to determine if the V912A mutation would be detrimental to DNMT3A protein function, and whether it was associated with patient 2's continuous complete remission (CCR) for 60 months. Both R885G and Q886R mutations were located in the methyltransferase domain and very close to hotspot R882. Thus, we speculated that these 2 mutations may play a role similar to that of mutation at R882, which is associated with poor prognosis. Mutation c.2739+55A>C which was located at 3 0 UTR changed the splice site of DNMT3A. However, the exact impact on the protein function and the association of this mutation with the poor prognosis of Pt.3 needed to be clarified in further functional experiments.
There have been few reports of DNMT3A mutations in patients with PML-RARA. A study has figured out that APL patients were generally excluded from many studies because of existence of the unique fusion gene PML-RARA and the favorable prognosis in contrast to other subtypes. Moreover, DNMT3A gene mutation was almost never found in APL www.md-journal.com Table 2 Clinicobiological characteristics and treatment outcome of childhood AML with  patients. [6,21] However, in our study, DNMT3A mutations were found in 2 patients (Pt.2 and Pt.4) with APL and PML-RARA fusion gene, indicating the requirement of detection of DNMT3A mutations in childhood APL. Now, more and more studies have shown that DNMT3A is a haploinsufficient tumor suppressor gene in myeloid leukemias, when cooperating mutations are present. [30][31][32] In the absence of high-risk cytogenetics, DNMT3A mutation status has an adverse impact on outcome in the presence of FLT3 and/or NPM1 mutations. [26] The R882H mutation associated with AML dominantly inhibits wild-type DNMT3A by blocking its ability to form active teramers. [33] DNMT3A loss drives enhancer hypomethylation in FLT3-ITD-associated leukemias. [34] Coexistence of DNMT3A R882 mutation and FLT3-ITD was an extremely poor prognostic factor in patients with normalkaryotype AML after allogeneic hematopoietic cell transplantation. [35] Notably in our study, the poor treatment outcome of Pt.4 in this study was consistent with the above results. Although both of Pt.2 and Pt.4 have PML-RARA fusion gene and mutated DNMT3A, the outcome of these 2 patients was quite different. Pt.4 (PML-RARA, DNMT3A Q886R , and FLT3-ITD positive) died at early phase of treatment, whereas Pt.2 (PML-RARA and DNMT3A V912A ) had been in CCR for 60 months, suggesting that FLT3-ITD acts as a key role to accelerate the progress in Pt.4 with APL and DNMT3A Q886R . DNMT3A mutations alone may not induce AML, only acts as an initial lesion and requires an additional genetic event to increase susceptibility to leukemic development.
The clinical significance of adult AML with DNMT3A mutations seems to be age dependent. [10,21] DNMT3A-R882 mutation are associated with adverse prognosis in older patients (≥60), and non-R882-DNMT3A mutations are associated with adverse prognosis in younger patients (18< and <60). This finding may also be available and suitable in childhood AML. In this study, the median age of the 4 patients with non-R882 DNMT3A mutations was 7 years old. In the 3 patients younger than 7, 2 patients (Pt.3 and Pt.4) died of relapse or complications, the other patient gave up treatment because of remission induction failure in day 33. In contrast, the only 1 patient older than 7 (Pt. 2) were in CCR up to 60 months. The mechanisms behind this phenomenon were worth investigating. Table 4 Association of DNMT3A mutation with common clinical characteristics.