From January 1999 to December 2008, a total of 99 newly diagnosed PML/RARA negative AML pediatric patients were included in this study. All 99 patients were already enrolled in the AML-XH-99 protocol and had available bone marrow (BM) samples frozen in the Biobank at Shanghai Children’s Medical Center.

At diagnosis, all patients underwent blood testing for complete blood counts and biochemistry panels. BM cells were aspirated for morphology, immunohistochemistry, karyotype, immunophenotyping, and molecular testings for RUNX1-RUNX1T1(AML1-ETO), CBFB-MYH11, or PML-RARA gene mutations (31). CN-AML is defined as leukemia without chromosomal abnormalities by karyotype and negative for RUNX1-RUNX1T1, CBFB-MYH11, or PML-RARA by PCR.

Additional BM samples were collected for research purposes at the time of diagnosis (after a legal guardian signed an informed consent). The mononuclear cells were isolated by Ficoll-Hypaque gradient and cryopreserved in the Biobank. This study was approved by the Institutional Review Board of Shanghai Children’s Medica Center.

All patients were treated on the AML-XH-99 protocol, which included induction, consolidation, and continuation therapies (30). Induction consisted two cycles of DAE (daunorubicin 40 mg/m²/day, days 1–3, cytarabine 200 mg/m² every 12 h, days 1–7, and etoposide 100 mg/m²/day, days 1–3). Consolidation included two courses of chemotherapy: (A) DA (cytarabine 2 g/m² every 12 h, days 1–3, and daunorubicin 30 mg/m²/day, day 1 and 2); and (B) EA (cytarabine 2 g/m² every 12 h, days 1–3, and VP-16 160 mg/m²/day, day 1 and 2). For continuation therapy, DA alternating with EA, followed by two courses of HA (homoharringtonine 3–4 mg/m²/day, days 1–9, and cytarabine 75 mg/m² every 12 h, days 1–7), were given for a total of 12 cycles (30).

Complete remission was defined as marrow blasts <5% with evidence of blood count recovery. Central nervous system leukemia prevention consisted of intrathecal (IT) chemotherapy with dexamethasone 5 mg/m², methotrexate 12.5 mg/m² (maximum 12.5 mg), and cytarabine 1 mg/kg (maximum 35 mg) at various time points (30). Allogeneic hematopoietic stem cell transplantation (Allo-HSCT) was performed in some relapsed or refractory patients with suitable donors who had achieved CR or at least a good partial response (PR, defined as a BM blast reduction of >50%) after salvage chemotherapy.

In this study, BM samples at diagnosis were analyzed for gene mutations. Control BM samples came from patients with idiopathic thrombocytopenia who underwent BM aspiration to rule out malignancy.

Total RNA was extracted from 1 × 10⁷ BM mononuclear cells (BMMC) using the phenol–chloroform–isopropanol method (TRIzol^®, Invitrogen). One microgram of total RNA was reverse transcribed according to the manufacturer’s protocols (Takara, Japan). RNA quality and quantity were assessed relative to the GAPDH housekeeping gene on the LightCycler^® 480 System (Roche Diagnostics, Penzberg, Germany). Samples with Ct_GAPDH values over 30 were considered uninterpretable.

The PCR and melting analysis for FLT3 and NPM1 mutations was performed on the LightCycler^® 480, a real-time PCR machine with HRM and 96/384 well capacity. All samples were tested in triplicates. One positive control and one negative control with each gene mutation, as well as one blank control with distilled water only, were included in triplets on each run of unknowns. PCR was performed from ~10 ng of cDNA, using 0.3 μM each of the relevant forward and reverse primers in a total volume of 10 μl containing 5 μl ES Taq HS premix (Takara) and 5 μM SYTO-82 (Invitrogen) (32). PCR was performed at 94°C for 2 min, followed by 45 cycles at 94°C for 10 s, 56°C for 10 s, 72°C for 20 s, and a final extension step at 72°C for 5 min. The cycling conditions were the same for all three amplicons, allowing them to be performed in one run. The program ran for 40 cycles at 95°C for 30 s, 40°C for 30 s, then 60°C −95°C (5 s, 1°C/s). Upon completion of the run, analysis was performed using the software supplied with the LightCycler^® 480. The melting curves were normalized, and temperature shifted to allow samples to be directly compared. Difference plots were generated by selecting a negative control as the baseline. The fluorescence of all other samples was plotted relative to this sample. Significant differences in fluorescence were indicative of mutations.

The primers used were: (1) FLT3-ITD: exon 11 (ITD-F) GCAATTTAGGTATGAAAGCCAGC, exon 15 (ITD-R) GTTGCGTTCATCACTTTTCCAA; (2) FLT3-TKD: TKD-F ACGTGCTTGTCACCC, TKD-R TTAATGGTGTAGATGCCTTCAAAC; and (3) NPM1: NPM-F TGACTGACCAAGAGGCTATTC, NPM-R GACAGCCAGATATCAACTGTTAC.

Real-time quantitative PCRs were performed on the LightCycler^® 480 using ~20 ng of cDNA and 0.5 μM each of the relevant forward and reverse primers: MLL-F3: AAGCAGCCTCCACCACCAGAAT and MLL-R3: CCACGAGGTTTTCGAGGACTA in a total volume of 10 μl containing with 5 μl SYBR Green I premix (Takara) at 94°C for 2 min, followed by 35 cycles at 94°C for 10 s, 56°C for 20 s. Samples with CT values below 35 and with satisfying melting peak at 82–84°C were considered as MLL-PTD mutants.

For wild-type samples, PCR products were purified and sent to outside facilities for Sanger sequencing. For samples with mutations, PCR products were purified and cloned into TA vectors (Takara, Japan). Plasmids extracted from clones that were successfully inserted with target genes were sent for Sanger sequencing.

Event-free survival was defined as the time elapsed from study enrollment to induction failure, relapse, or death. Overall survival (OS) was defined as the time from study enrollment until death from any cause. EFS and OS were calculated using the Kaplan–Meier method, and the log-rank test was used for comparisons of Kaplan–Meier curves. All p-values were two-sided and p < 0.05 was considered as statistically significant. Statistical calculations were performed using SAS version 9.2 (SAS Institute Inc., Cary, NC, USA).

Results were analyzed using the LightCycler^® 480 associated software (release 1.5.0). Data were presented in two formats: the normalized plot (Figures 1A–C), in which the amount of intercalating dye remaining at any temperature point was expressed as a fraction of the amount prior to data acquisition; and a difference plot (Figures 1D–F), where the average HRM profile of the control samples was used by the genotype function of the machine software as the standard wild-type profile for subsequent comparison to each of the test samples. Each mutant allele had its own distinctive melting curve when compared to the wild-type allele. The distinct melting curves of the mutant became more apparent when data were represented in a difference plot format than in a normalized plot.

To assess the specificity of the HRM test, we selected nine samples, seven of which tested positive and two negative for FLT3-ITD. Direct Sanger sequencing confirmed 100% consistency. Furthermore, we performed similar validation for the NPM1 gene mutation in six mutants and five wild-type samples. Results were confirmed with 100% consistency when compared to Sanger sequencing (Table 1). Sanger sequencing verification showed that in patients with FLT3-ITD, the size of internal tandem duplication varied from 3 to 66 bp. No difference was observed between the size of the ITD and the results seen in HRM analysis in all the samples analyzed. Sequencing of samples with FLT3-TKD confirmed the most frequent D835Y mutation. The NPM1 mutation involved 4 bp insertions, which altered the tryptophan at amino acid position 288.

We also tested whether HRM analysis could be used as a method for MRD monitoring. The ability to detect low levels of FLT3-ITD, FLT3-TKD, and NPM1 gene mutations in a background of non-mutated DNA was evaluated by titrating each of the mutant alleles with wild-type DNA to produce a range of the mutant allele dilutions. Unfortunately, the lower limit of detection was 1/10 ~ 1/100 for FLT3-ITD and 1/100 ~ 1/1000 for both FLT3-TKD and NPM1. This suggested that HRM analysis was not sensitive enough for MRD monitoring in this study (Figure 2).

Patient characteristics were summarized in Table 2. The mean age of patients was 7 years (range 0.3–16.9 years). Slightly more males than females were included in the study. Five patients had M3 morphology. However, they were all negative for t(15;17) and PML-RARA. Therefore, they were all treated with AML-XH-99 protocol and were included in the analysis. Among all 99 PML-RARA negative AML patients, 22 (24.2%) were positive for RUNX1-RUNX1T1, and 1 (1.0%) was positive for CBFB–MYH11. Fifty four patients (54.5%) had normal karyotype and were defined as CN-AML.

Of the total 99 patients, HRM analysis detected 33 patients with FLT3-ITD mutations, 10 with FLT3-TKD mutations, and 21 with NPM1 mutations. Four patients were positive for MLL-PTD mutations (Table 3) by RQ-PCR. The presence of the mutations was not mutually exclusive. Among the 33 patients with the FLT3-ITD mutations, 4 had NPM1 mutations, and 2 patients had the MLL-PTD mutations. Similarly, among the 10 patients with the FLT3-TKD mutations, 2 of them had NPM1 mutations and 1 had the MLL-PTD mutation. One patient was found to have concurrent FLT3-ITD, FLT3-TKD, and NPM1 mutations. In total, 66% patients had at least one gene mutation.

Of the 54 CN-AML patients, there were 18 (33.3%) FLT3-ITD mutants, 4 (7.4%) FLT3-TKD mutants, 13 (24.1%) NPM1 mutants, and 3 (5.56%) MLL-PTD mutants (Table 3).

The EFS and OS at the median follow-up time of 49.6 months for all patients were 45.2 and 50.0%, respectively. The EFS of patients with FLT3-ITD mutations were significantly worse when compared to that of patients without FLT3-ITD mutations (p = 0.038, Figure 3A). The OS for patients with and without FLT3-ITD mutations were 38 and 55%, respectively. However, the difference was not statistically significant (Figure 3D). There was no significant difference in EFS and OS for patients with or without FLT3-TKD mutations (Figures 3B,E). The EFS and OS for all patients with NPM1 mutations were significantly better than patients without NPM1 mutations (p = 0.01, Figures 3C,F). All four patients with MLL-PTD mutations died from disease.

In the CN-AML group of 54 patients, the EFS and OS time were 50 and 53%, respectively. We found no significant difference in EFS and OS for patients with or without FLT3-ITD mutations (Figures S1A,D in Supplementary Material) or FLT3-TKD mutations (Figures S1B,E in Supplementary Material). There is a trend toward better EFS in patients with the NPM1 mutation when compared to patients without the mutation (p = 0.07, Figure S1C in Supplementary Material); OS was significantly better (p = 0.05, Figure S1F in Supplementary Material). All three patients with MLL-PTD mutations in the CN-AML group died of disease progression.