Spectrum and Clinical Characteristics of Gene Mutations in Chinese Pediatric Acute Lymphoblastic Leukemia

Purpose The 5-year survival rate of children with acute lymphoblastic leukemia (ALL) is 85–90%, with a 10–15% rate of treatment failure. Next-generation sequencing (NGS) identied recurrent mutated genes in ALL that might alter the diagnosis, classication, prognostic stratication, treatment, and response to ALL. Few studies on gene mutations in Chinese pediatric ALL have been identied. Thus, an in-depth understanding of the biological characteristics of these patients is essential.. The present study aimed to characterize the spectrum and clinical features of recurrent driver gene mutations in a single-center cohort of Chinese pediatric ALL. Methods We enrolled 219 patients with pediatric ALL in our single center. Targeted sequencing based on NGS was used to detect gene mutations in patients. The correlation was analyzed between gene mutation and clinical features, including patient characteristics, cytogenetics, genetic subtypes, risk stratication and treatment outcomes using χ 2 -square test or Fisher’s exact test for categorical variables.


Introduction
Pediatric acute lymphoblastic leukemia (ALL) is the most common childhood malignancy with 5-year overall survival (OS) rate of 85-90% and treatment failure rate of 10-15% (Dores et al. 2012; Inaba et al. 2013). Next-generation sequencing (NGS) identi ed recurrent mutated genes in pediatric ALL that might alter the diagnosis, classi cation, prognostic strati cation, treatment, and response to ALL. However, there are few studies on gene mutations in Chinese pediatric ALL. Thus, an in-depth understanding of the biological characteristics of these patients is essential, and it is necessary to conduct comprehensive and thorough gene mutation detection by NGS in ALL patients. In the current study, we aimed to characterize the spectrum and clinical features of gene mutations in a single-center cohort of Chinese pediatric ALL patients.

Patients
The cohort in the present study consisted of 219 children (0.05-16.25, median: 3.75 years) with newly diagnosed ALL (n=196, B-cell ALL (B-ALL); n=23, T-cell ALL (T-ALL)) at our hospital between October 2017 and October 2019. The protocol was approved by the Medical Ethics Committee of the Children's Hospital, and written informed consent was obtained from the parents or guardians. ALL was diagnosed based on the morphology, immunophenotyping, cytogenetics, and the molecular biology of leukemia cells (Sabattini et al. 2010). Flow cytometric (FCM) immunophenotyping of bone marrow was performed on FACSCalibur with CellQuest software. A total of 51 fusion genes, including ETV6-RUNX1, BCR-ABL1, and MLL rearrangement, were examined by polymerase chain reaction (PCR). Karyotyping analysis was conducted by conventional methods. The patients were according to the National Protocol of Childhood Leukemia in China (NPCLC)-ALL2008 protocol, a modi ed form of protocol NPCAC97 (Tang et al. 2008).
Next-generation sequencing and mutation analysis Genomic DNA was extracted from bone marrow samples at diagnosis. The spectrum of gene mutations was determined through NGS platform at Acornmed Biotechnology Co. Ltd (Beijing, China). Genetic pro ling included the targeted sequencing of 185 genes (Supplementary Table S1). Multiplex libraries were sequenced using Illumina NovaSeq. The following criteria were used to lter raw variant results: average effective sequencing depth on target per sample ≥1,000x; variant allele frequency (VAF) ≥1% for single nucleotide variations (SNVs), insertions, or deletions (InDels); mapping quality ≥30; and base quality ≥30. The reads were aligned to the human genome using Burrows-Wheeler alignment (BWA, version 0.7.12). PCR duplicates were removed using the MarkDuplicates tool in Picard. Genome Analysis Toolkit (GATK; version 3.8) comprising of IndelRealigner and BaseRecalibrator was applied for realignment and recalibration of the BWA data, respectively. Mutect2 was used to identify SNVs and InDels. All the variants were annotated by ANNOVAR software, including 1000G projects, COSMIC, SIFT, and PolyPhen.

Statistics
The correlations between various gene mutations and clinical features were analyzed using χ 2 -square test or Fisher's exact test for categorical variables A two-sided P value <0.05 was considered to indicate statistical signi cance.
Conversely, the rest of the mutated genes with lower VAF < 50% indicated that they were present in a subpopulation of the sequenced cells.. Notably, four RAS signaling pathway mutated genes (FLT3, NRAS, KRAS, and PTPN11) had minimal median VAFs, suggesting that these genes were subclones.
Next, we investigated the co-occurrence of mutated genes (Fig. 4A) and found signi cant associations between mutated NOTCH1 and mutations in FBXW7 and PTEN, mutated JAK2 and mutations in MSH6 and PCLO, and mutated DNM2 and mutations in PHF6 and USP7. Moreover, pairwise associations were observed between ETV6 and KRAS, KDM6A and KMT2D, RUNX1 and ATRX, ASXL1 and SH2B3, and USP7 and FBXW7 (p < 0.05, Fig. 4B).

Correlation Between Gene Mutations And Patient Characteristics, Cytogenetics
The correlation analysis between gene mutations and the patient characteristics in these patients revealed that PIK3R1 mutation was more common in infants compared to patients ≥ 1-year-old (p = 0.021, Fig. 5A). Moreover, KMT2D was more common, while PTPN11 mutation was less common in patients > 3.75-year-old (Fig. 5B). We also found that gender did not in uence the mutational status of any of the genes. In addition, the mutations in NOTCH1 and PTEN were more common in patients with initial leukocyte count > 50×10 9 /L (P < 0.0001 and P = 0.097, respectively) (Fig. 5C). Patients with FLT3 mutations showed lower platelet counts (≤ 62 ×10 9 /L, P = 0.024, Fig. 5D) and hemoglobin level (≤ 82 g/L, P = 0.0499, Fig. 5E) at diagnosis than those without FLT3 mutations. While patients with NOTCH1 mutations had a high hemoglobin level (> 82 g/L, P = 0.0054) (Fig. 5E). Compared to B-ALL, NOTCH1, PTEN, FBXW7, USP7, DNM2, and CDKN2A were more frequently mutated in T-ALL (all P < 0.05) (Fig. 5F). Futhermore, KRAS and FLT3 mutations were both enriched in patients with hyperdiploidy (P < 0.0001, P = 0.0003, respectively, Fig. 5G).
Correlation between gene mutations and genetic subtypes, risk strati cation and treatment outcomes Molecular genetic analyses of 51 fusion transcripts, including ETV6-RUNX1, TCF3-PBX1, BCR-ABL1, and KMT2A (MLL) rearrangement were conducted successfully in all the patients. Strikingly, NRAS, PTPN11, FLT3, and KMT2D mutations were common in patients who did not carry the fusion genes (all p < 0.05), and NRAS mutations were rarely in patients with ETV6-RUNX1 (P = 0.002). RUNX1 and ROBO1 mutations were more closely linked to BCR-ABL1 fusion gene (P = 0.017, P = 0.032, respectively), and PAX5, PHF6, and STAG2 mutations were associated with TCF3-PBX1 fusion gene (P = 0.001, P = 0.006, P = 0.046, respectively). MLL translocations co-existed with PIK3R1 mutation (P = 0.017, Fig. 6A). Some genes are closely associated with the prognosis of the disease. Herein, we studied the associations of gene mutations with risk strati cation in the cohort. PTEN mutation was signi cantly associated with high risk ALL patients (P = 0.011), while NOTCH1 mutation was common in intermediate risk ALL patients (P = 0.039, Fig. 6B). In addition, PIK3R1 mutation occurs frequently in high risk B-ALL patients (P = 0.023, Fig. 6C). Patients with ETV6 or PHF6 mutations detected at the time of diagnosis were more sensitive to steroid treatment (P = 0.033, P = 0.048, respectively, Fig. 6D). In B-ALL patients, we analyzed the associations between gene mutations and early MRD levels (MRD1 on day 15 and MRD2 on day 33).

Discussion
In this study, we dissected the genetic landscape, analyzed the mutational spectrum of various immunological ALL lineages, and explored the correlations between mutational and clinical features, Although our study showed that T-ALL had a signi cantly higher mutation level than B-ALL, the average number of mutations was still lower than the expected value. This deviation could be attributed to the scope of sequencing, the evaluated variation types, the sensitivity of the test, and the lter criteria of mutation calling. Consistent with the ndings of previous studies, several genes, such as DNM2, PHF6, WT1, and RPL10 were found to be involved in increased kinase signaling, transcription factors, epigenetic factors, translation, and RNA stability at a low frequency in our cohort (Belver and Ferrando 2016; Girardi et al. 2017). We also found that the accumulation of mutations in T-ALL did not occur randomly (Vogelstein et al. 2013). Interestingly, the coexistence of NOTCH1-PTEN-FBXW7 and DNM2-USP7-PHF6 mutations was observed in our T-ALL cohort. The coexistence phenomenon suggested that those Notch pathway and non-Notch pathway genes interconnect physiologically and cooperate during the development and progression of the T-ALL, respectively.
MLL translocations and PIK3R1 mutations were common in infant ALL, a group characterized as immature cytologically, resistant to conventional therapies, and showing poor prognosis. In addition, the MLL gene arrangement is the hallmark of infant leukemia, associated with a high incidence (approximately 80%) ( However, our study also has some limitations. First, the enrolled patients were from our single center, which could not well re ect the gene mutation level of the whole Chinese pediatric ALL, and further multicenter studies are needed. Second, long-term follow-up of patients is needed to explore the relationship between mutations and prognosis.
In summary, our study depicted the speci c genomic landscape and revealed the relevance between mutational spectrum and clinical features of Chinese pediatric ALL in a single cohort, including patient characteristics, cytogenetics, genetic subtypes, risk strati cation and treatment outcomes. The discovery of this mutational spectrum highlights the need for molecular classi cation, risk strati cation, and prognosis evaluation and also provide the basis for the development and application of new targeted therapy for pediatric ALL.