This is a retrospective, descriptive clinical study involving pediatric patients diagnosed with B-ALL at a public pediatric oncology referral hospital in Brasília, Brazil. A convenience sampling was performed, including patients admitted between July 2018 and September 2023. Data collected comprised clinical characteristics (age, gender, white blood cell count at diagnosis, and the presence of Central Nervous System—CNS—infiltration), biological characterization (cytogenetic alterations, recurrent rearrangements, and mutations in IKZF1, PAX5, and ERG), and treatment response measured by Minimal Residual Disease (MRD). Outcomes were categorized as remission, relapse or dead in remission.

Patients aged 1–18 years with a primary diagnosis of B-ALL and no prior treatment were included. Between July 2018 and September 2022, treatment was based on the BFM ALLIC 2009 protocol, locally adapted for B-ALL management at our institution. Since September 2022, the GBTLI 2021 protocol—a Brazilian multicenter research protocol in which our institution participates—has been used. As a result, patients in this study followed two different protocols. The backbone strategy for both protocols is similar, though there are differences in the criteria used for risk classification. The GBTLI 2021 protocol reduces the intensity of induction therapy for patients classified as low and intermediate risk (Supplementary Frames S1, S2, and S3).

Bone marrow samples were collected via aspiration to confirm ALL diagnosis. Immediately after collection, smears were prepared for morphological evaluation. Bone marrow aspirate samples (or peripheral blood in cases of high white blood cell counts and patient severity) were sent to the Translational Research Laboratory for immunophenotyping by flow cytometry, cytogenetics (cell culture), and molecular biology analysis.

MRD was evaluated by flow cytometry according to Euroflow Consortium guidelines (50). Cellular events were acquired using the FACS Canto II cytometer (BD), with data analysis performed using FACS Diva (BD) and Infinicyt (Cytognos, version 2.0) software. MRD values were measured at the mid-point and conclusion of the induction phase. For patients treated with the adapted ALLIC BFM 2009 protocol, MRD was assessed on days 15, 33, and 78. For patients following the GBTLI 2021 protocol, MRD quantification was conducted on days 19, 26 (for the low-risk subgroup), and 49.

Mononuclear cells were isolated using a Ficoll gradient (GE Healthcare Life Sciences), washed in 1X PBS, and aliquoted into two aliquots. One tube was used for total RNA extraction via the Trizol® method (Invitrogen), and the other for DNA extraction. DNA was extracted using the Wizard Genomic DNA Purification Kit (Promega) following manufacturer instructions. Samples were quantified via spectrophotometry and stored at −20°C (for DNA) and −80°C (for RNA) until assays were performed.

Screening for genetic variants in the FLT3 tyrosine kinase domain (D835) was conducted on all patients with available samples (n = 155/156) via restriction fragment length polymorphism (RFLP) analysis (51). The PCR reaction included 1x PCR buffer, 200 nM each dNTP, 1.5 mM MgCl₂, 0.75 U Platinum™ Taq DNA Polymerase (Thermo Fisher Scientific), and 0.2 mM of each primer (FWD-CCGCCAGGAACGTGCTTG, REV-CAGCCTCACATTGCCCC). PCR conditions were as follows: 95°C for 3 min, 35 cycles of 95°C for 30 s, 56°C for 30 s, and 72°C for 1 min, followed by a final extension at 72°C for 5 min. PCR products were digested using the restriction enzyme EcoRV (NEB), and samples showing alterations by RFLP were subjected to Sanger sequencing.

Detection of genetic variants in the juxtamembrane domain was performed by fragment analysis (51) on the ABI3500 Genetic Analyzer (Applied Biosystems). PCR was performed under the same conditions described above, but with 27 cycles, using FWD-6-FAM-GCAATTTAGGTATGAAAGCCAGC and REV-CTTTCAGCATTTTGACGGCAACC primers. Fragment size was estimated using the GeneScan™ 500 LIZ size standard (Thermo Fisher Scientific).

For the classification of somatic variant pathogenicity in cancer (oncogenicity), we applied the Standard Operating Procedure developed in accordance with recommendations from the Clinical Genome Resource (ClinGen), the Cancer Genomics Consortium (CGC), and the Variant Interpretation for Cancer Consortium (VICC) (52).

FLT3 expression was analyzed in patients admitted between July 2018 and December 2022 with available samples (n = 112/130) using real-time RT-PCR (RT-qPCR) on bone marrow samples collected at diagnosis. FLT3 expression was also evaluated in in 10 samples collected at relapse. One microgram of RNA was used for cDNA synthesis, followed by PCR with 200 nM primers and 2x PCR MasterMix containing SybrGreen®. Relative quantification was performed using the 2-ΔΔCq method, with HPRT1 and B2M as reference genes. The calibrator was the median ΔCq from all cases in the study, excluding relapses. Singleplex reactions were conducted on a QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific) with primers for FLT3 (FWD-AGGGACAGTGTACGAAGCTG; REV-GTCGTGCTTAAAGACCCAGAG), HPRT1 (FWD-TGACACTGGCAAAACAATGCA; REV-GGTCCTTTTCACCAGCAAGCT), and B2M (FWD-TGCTGTCTCCATGTTTGATGTATCT; REV-TCTCTGCTCCCCACCTCTAAGT). Amplification conditions were: 2 min at 50°C, 10 min at 95°C for the holding stage, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Melt curve analysis was performed to assess amplicon specificity.

Data were tested for normal distribution, by the D'Agostino and Pearson normality test and analyses of skewness and kurtosis, when applicable. Data were expressed as average ± SD (qPCR), mean ± SEM or median and ranges according to the distribution. Statistical analysis was performed using GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego California USA). The statistical approach adopted for each analysis is described in the figure legends. Non-parametric tests were used for data with non-normal distribution and probability values of p < 0.05 were accepted as indication of statistically significant difference.

The study was approved by the local research ethics committee (protocol code 44796221.9.0000.0144. July 04, 2021). Informed consent was obtained from guardians, and assent forms were signed by patients over 5 years old.

A total of 208 patients diagnosed with B-ALL were treated at our institution from July 2018–September 2023, with 156 meeting the inclusion criteria for this study (Figure 1). Patient characterization included age, sex, CNS status, treatment protocol applied, initial and post-induction risk classification, MRD values at mid-induction and post-induction, disease progression and current clinical status, cytogenetic alterations, recurrent rearrangements, presence of iAMP21, IKZF1 deletions (whether associated with the IKZF1 Plus subtype or not), PAX5 alterations, P2RY8::CRLF2 rearrangement, FLT3 mutations, and FLT3 expression levels.

The patients' ages ranged from 1–17 years, with a peak incidence between ages 2 and 5, and a median age of 4 years and 5 months. Thirty-four patients (21.8%) were over 10 years old at admission, and the sex ratio was 1.05 (80 males to 76 females). Regarding white blood cell counts at admission, 133 patients (85.2%) presented counts below 50,000 leukocytes/mm³, while 33 patients had counts above 50,000/mm³, including 12 with counts exceeding 100,000/mm³.

Most children (122) received treatment following the ALLIC BFM 2009 protocol. In terms of risk classification at the end of induction, 25 patients (20.4%) were classified as low risk, 74 (60.6%) as intermediate risk, and 23 (18.8%) as high risk. Among those treated under the GBTLI 2021 protocol, risk classifications were as follows: 16 (47%) intermediate risk, 12 (33%) high risk, 4 low risk, and 2 very high risk (Supplementary Table S1). Differences in risk classification criteria and treatment intensity between the two protocols led to a higher frequency of “high-risk” patients under the GBTLI protocol; however, this difference does not imply greater disease aggressiveness in this group.

The 5-year overall survival (OS) and event-free survival (EFS) rates for this cohort were 87.5% and 78%, respectively, based on a minimum 5-year follow-up for 24 patients from the time of diagnosis. The clinical and laboratory data of the patients are presented in Figure 2.

FLT3 mutation screening was conducted in 155 patients (Figure 1), revealing nine mutations in eight patients (5.1%). One type of mutation in the tyrosine kinase domain (p.Ile836del) was identified in two patients. Four patients (2.58%) presented with FLT3-TKD mutations, three patients (1.93%) had in-frame insertions and deletions in the juxtamembrane domain (FLT3-JM-INDEL), and one patient (0.64%) had mutations in both the tyrosine kinase domain (FLT3-TKD) and the juxtamembrane domain (FLT3-JM-INDEL) (Figure 3). No FLT3-ITD variants were found in our cohort.

The four mutations identified in the tyrosine kinase domain had been previously documented. Among these, three mutations have been associated with gain of function and oncogenic potential, while one mutation is likely to exhibit oncogenic potential due to a probable gain of function. None of the four mutations identified in the juxtamembrane region had been reported prior to this study. These alterations demonstrated moderate oncogenic potential attributed to a possible gain of function (Table 1).

No association was found between FLT3 mutations and age, gender, white blood cell (WBC) count at diagnosis, or minimal residual disease (MRD) values measured at the mid-point and at the end of induction therapy. Table 2 presents the clinical features and detailed descriptions of the FLT3 mutations.

FLT3 mutations were associated with high hyperdiploidy in 3 out of 8 patients (37.5%) (patients P19, P104, P112,). All three cases had TKD-type mutations, and one patient presented with a JM-INDEL associated with a TKD mutation (P112). No FLT3 mutations were identified in any patients with ETV6::RUNX1 (n = 38), BCR::ABL1 (n = 4), TCF3::PBX1 (n = 9), or r-KMT2A (n = 2). Additionally, none of the patients with P2RY8::CRLF2 (n = 7), IKZF1 deletions (n = 21), including IKZF1plus (n = 11), PAX5 alterations (n = 21), or ERG deletions (n = 9) presented with FLT3 mutations (Table 2, Figure 2).

Among patients with FLT3 mutations, a relapse rate of 25% (2/8) was observed, with both cases occurring in patients with the FLT3-TKD mutation and no associated high hyperdiploidy. Although this difference was not statistically significant compared to FLT3 wild-type (WT) patients (p = 0.08), it is noteworthy given the rarity of the mutation. The relapse rate among FLT3 WT patients was 5.4% (8/147) (Figure 4).

Follow-up of the two relapsed patients with FLT3-TKD mutations revealed that one experienced a very early relapse, failed to achieve remission, and died three months after the relapse diagnosis (FLT3 mutation screening was not conducted on the relapse sample). The other patient achieved remission but experienced a second relapse with an orbital lesion. Treatment included chemotherapy and orbital radiotherapy (20 Gy). The mutation in FLT3 identified in the initial diagnosis sample was also detected in the bone marrow sample collected at the time of relapse. This patient is currently alive, with a follow-up period of 55 months.

There was no difference in survival rates when comparing patients with FLT3 mutations to those with FLT3 WT (Figure 5).

FLT3 expression was evaluated in RNA samples of 112 patients at diagnosis. Additionally, FLT3 expression was assessed in 10 samples collected at relapse.

The distribution of relative FLT3 expression in the samples showed a non-Gaussian pattern, with a strong rightward skew (Figure 6A), indicating high expression levels without correspondingly low levels, as would be expected for independent variables with a normal distribution. This pattern is typical of genes with biological significance in carcinogenesis, where elevated expression levels confer an advantage, and lower levels are incompatible with disease maintenance (54, 55).

In 93 patients, the expression ratio (RQ) ranged from 0.017–3. Fifteen patients had expression ratios between 3 and 6.13, and in four patients, the expression ratio exceeded 10, which we classified as “hyperexpression.” These elevated RQ values were observed exclusively in four wild-type (WT) FLT3 patients. We were unable to identify factors associated with FLT3 hyperexpression in these four patients with RQ > 10 (see Table 3), but we observed that these individuals demonstrated a rapid response to treatment, as indicated by MRD values assessed during mid-induction (D15 or D19, depending on the protocol) (Figure 6B). Conversely, there was no statistically significant difference in MRD values at the end of induction between patients with hyperexpression and those without (Figure 6C).

Although none of the patients with FLT3 mutations exhibited hyperexpression of the gene (RQ > 10), FLT3 expression values were higher among patients with FLT3 mutations compared to wild-type patients (median 2.53 vs. 1.03, p = 0.03) (Figure 6D).

No significant differences in FLT3 expression values were observed among patients across different risk groups (data not shown).

A noteworthy finding is that FLT3 expression levels were significantly lower in relapse cases. Patients who relapsed had relative expression levels (at initial diagnosis) comparable to the group median (Figure 7A). However, these values decreased markedly after relapse (Figure 7B). This difference is even more pronounced when comparing paired expression values at diagnosis and at relapse for each patient (Figure 7C). Despite this, FLT3 expression levels had no significant impact on survival rates (EFS and OS) (Figure 8, Supplementary Table S2).

In our study, we found FLT3 mutations in 5.1% of the samples (8/155), with 3.2% (5/155) of mutations occurring in the TKD domain (in one case, associated with JM-INDEL). Unlike acute myeloid leukemia (AML), the variants found in the juxtamembrane domain were indels rather than internal tandem duplications (ITDs).

These findings are consistent with those reported in previous studies regarding both the frequency of mutations and the different spectrum of FLT3 mutations in B-cell acute lymphoblastic leukemia (B-ALL) compared to AML (20, 21). Taketani et al. (2004) found FLT3-TKD mutations in 6 (5.4%) of 112 children with ALL older than 1 year and in 8 (16.0%) of 50 infants with ALL, but no FLT3-ITD mutations were detected (32). Zhang et al. (2020) reported a predominance of JM-INDELs in B-ALLs (21). In Saudi Arabia, 4.7% of children with B-ALL had FLT3 mutations (2.4% ITD) (33), and in Brazil, Barbosa et al. found FLT3 mutations in 6.7% of 134 B-ALL patients (34). Additionally, this frequency was reported to be 5.5% among Canadian children and adolescents with ALL, including 1.1% FLT3-ITD and 4.3% FLT3-TKD point mutations, with an observed association between FLT3 mutations and hyperdiploidy (36).

We recognize that broader horizontal coverage and the ability to simultaneously detect multiple genetic alterations [single nucleotide variants [SNVs], copy number variations [CNVs], fusions, and indels] using highly sensitive techniques like next-generation sequencing (NGS) allow for the identification of FLT3 mutations at higher frequencies than restriction fragment length polymorphism (RFLP) and fragment analysis (20, 44). While NGS provides comprehensive genomic insights, it is time-consuming and may produce false negatives due to amplification issues or software limitations in detecting FLT3 internal tandem duplications (FLT3-ITD). In contrast, PCR fragment analysis and PCR-RFLP targeting FLT3-TKD (D835/I836) mutations are robust, cost-effective methods that deliver faster results, making them a viable and efficient alternative to NGS (57, 58).

In a recently published study, Zhao et al. described a higher frequency of FLT3 mutations in ALL patients when evaluated by NGS (6.3% ITD and 18.8% TKD) and identified new non-canonical genetic variants, such as point mutations outside the TKD and insertion/deletion variants causing in-frame amino acid alterations. In the same study, a higher proportion of patients with negative minimal residual disease (MRD) at mid-induction (D19) was observed among patients with FLT3 mutations compared to wild-type patients (20). In our sample, we did not identify differences in MRD levels between patients with or without FLT3 mutations.

Regarding initial presentation, unlike what is well established in patients with AML, where FLT3 mutations are associated with elevated leukocyte counts at diagnosis (22, 23, 59), none of the patients with FLT3 mutations in our study presented with hyperleukocytosis. Furthermore, no patients with FLT3 mutations had central nervous system (CNS) involvement at diagnosis.

In our sample, FLT3 mutations were not detected in patients with recurrent genetic alterations commonly associated with B-ALL, including ETV6::RUNX1, TCF3::PBX1, BCR::ABL1, r-KMT2A, P2RY8::CRLF2 rearrangements, PAX5 alterations, or deletions involving IKZF1. This finding supports the hypothesis that, contrary to the prevailing concept of kinase alterations being secondary (31, 60), FLT3 mutations may function as leukemogenic drivers in a small subset of B-ALL (5, 61). Further studies utilizing bone marrow samples collected during treatment monitoring or at relapse could clarify the role of FLT3 mutations in disease progression. Preliminary studies suggest that FLT3 mutations negatively impact prognosis in infants with r-KMT2A B-ALL (31, 49). However, in patients with hyperdiploidy, the presence of FLT3 mutations did not affect the natural course of the disease in this subgroup, which generally has a favorable outcome (31–33). In our study, among the five patients with TKD mutations, three who also had hyperdiploidy achieved remission, while the other two without hyperdiploidy experienced relapse. In light of these results and based on evidence of an association between hyperdiploidy and higher levels of FLT3 expression, as well as previous studies linking elevated FLT3 expression with certain subtypes of B-ALL (r-KMT2A, Ph-like), we decided to complement our research by evaluating FLT3 expression levels.

Higher FLT3 expression levels are described and considered recurrent alterations in acute leukemias. The expression levels of this gene are higher in cells from patients with acute leukemias compared to normal bone marrow samples and other types of neoplasms (37–43). In the context of B-ALL, subgroups of patients with r-KMT2A, high hyperdiploidy, and r-ZNF384 subtypes exhibit higher FLT3 expression levels (5, 36, 37, 40, 42). In our study, which was limited to patients older than one year, only two patients had r-KMT2A, both with expression values above the 75th percentile of our cohort.

Although the association between hyperdiploidy and higher levels of FLT3 expression has been documented in some studies (44, 61, 62), we did not find differences in FLT3 expression in this group. Similarly, there was no correlation between FLT3 expression and indicators of aggressiveness in B-ALL, such as white blood cell count at diagnosis or CNS involvement. This lack of association may be specific to our study population; however, expanding the sample size is necessary to confirm these findings.

Additionally, we were unable to identify any common characteristics among the four patients with extremely high expression levels (RQ > 10). Yang and colleagues demonstrated that epigenetic alterations with enhancer hijacking secondary to the deletion of the PAN3 gene (13q12.2) explain elevated FLT3 expression in B-ALL patients, particularly among those with hyperdiploidy or those who experienced relapse. Although none of the four patients with RQ > 10 had associated hyperdiploidy or experienced disease relapse, this could be a possible mechanism to explore in these cases (42).

The implications of FLT3 expression levels on prognosis remain uncertain and controversial. Among patients with r-KMT2A B-ALL, high FLT3 expression levels have been associated with poorer outcomes (38, 40, 49). However, in 2017, Fedders and colleagues found an opposite association (63). In our study, we were unable to identify any influence on survival rates in patients with B-ALL and FLT3 overexpression detected in bone marrow samples at diagnosis. The majority of authors also did not find an association between FLT3 expression levels at diagnosis and survival or relapse rates (20, 36). On the other hand, Garza Veloz reported that high FLT3 expression levels at the end of induction were associated with higher relapse and mortality rates (40).

We identified evidence that FLT3 may be a marker influencing the biological behavior of B-ALL, reinforcing the need for more in-depth studies to better elucidate FLT3's relationship with mechanisms involved in the genesis, survival, or resistance of leukemic cells to chemotherapeutic effects: 1. the frequency distribution of relative FLT3 expression (Figure 6); 2. the low MRD values observed during mid-induction in patients with high FLT3 expression (Figure 6); and 3. significantly lower FLT3 expression levels in samples from relapsed patients (Figure 7).

Given these results, the detailed biological effects and prognostic impact of FLT3 expression levels should be further investigated, particularly in the context of the potential use of FLT3 inhibitors. Additionally, it is important to not only focus on transcript expression levels but also evaluate the true functional impact of FLT3 by assessing receptor saturation or activation.

The small sample size, particularly given the low frequency of FLT3 mutations, along with the limited follow-up time for the patients, constrains the interpretation of our findings and underscores the necessity for continued research in this area. We acknowledge the inherent limitations of Restriction Fragment Length Polymorphism (RFLP) and fragment analysis in detecting low-frequency FLT3 mutations, especially when compared to the enhanced sensitivity of Next-Generation Sequencing (NGS).

Nevertheless, considering the potential for direct and short-term benefits for selected patients, we believe it is crucial to disseminate our findings, even with the limited follow-up period. In light of these limitations, we assert that our findings contribute meaningfully to the understanding of FLT3 mutation prevalence and advocate for further research and investment in comprehensive molecular diagnostics in similar contexts.

Preclinical studies and case reports provide evidence supporting the use of FLT3 inhibitors in specific subgroups of B-ALL patients (3, 45, 46, 56). However, although the use of FLT3 inhibitors has been suggested as a therapeutic option for B-ALL patients for nearly two decades, studies exploring this approach remain scarce (46, 56).

We propose that in selected cases, alongside MRD assessment, evaluating FLT3 status (mutation or expression) may serve as an additional tool in guiding treatment strategies. For instance, in patients with persistent MRD following induction or in relapse cases with FLT3 mutations or increased expression, FLT3 inhibitors could be considered as salvage treatment. Another option is to use FLT3 inhibitors in combination with conventional chemotherapy to reduce the required doses in patients with a favorable prognosis. FLT3 inhibitors may also be beneficial for patients with FLT3 mutation or overexpression where high-dose chemotherapy is contraindicated, such as in cases of infection or toxicity.