Prognostic significance of copy number variation in B-cell acute lymphoblastic leukemia

Copy number variations (CNVs) are widespread in both pediatric and adult cases of B-cell acute lymphoblastic leukemia (B-ALL); however, their clinical significance remains unclear. This review primarily discusses the most prevalent CNVs in B-ALL to elucidate their clinical value and further personalized management of this population. The discovery of the molecular mechanism of gene deletion and the development of targeted drugs will further enhance the clinical prognosis of B-ALL.


Introduction
B cell acute lymphoblastic leukemia (B-ALL) is a heterogeneous and invasive hematological malignancy with the accretion of genetic lesions (1,2). Recent research has comprehensively investigated the genetic landscape of both adult and pediatric B-ALL (3)(4)(5). Over 90% of pediatric patients with B-ALL can attain complete remission (CR), 20% relapse, and 10% remain incurable (6). The conventional approach for pediatric B-ALL remission-induction chemotherapy drugs mainly consists of glucocorticoid, vincristine, asparaginase and/or anthracycline (7). With the first course of induction therapy administration for 4-6 weeks, the CR rate population of pediatric B-ALL may reach 98% (7).
The genomic pattern of adult B-ALL might differ from pediatric cases, accompanied by more devastating clinical outcomes (8). However, chances of newly emerged drugs, chimeric antigen receptor T cell therapy, and hematopoietic stem cell transplantation (HSCT) improved the clinical response of specific subtypes of B-ALL patients remarkably (9)(10)(11)(12). Nevertheless, 40% of adult patients with B-ALL relapsed at a median duration of 13 months (28 days to 12 years) (13). In this population, around 30%-40% of relapsed and refractory B-ALL cases can attain complete remission by first salvage chemotherapy. Besides, the long-term survival, that is, the 5-year survival rate, of patients with B-ALL remains at 20% only (14,15). Hence, it is imperative to find a n o v e l b i o m a r k e r t h a t c o u l d h e l p d e t e r m i n e t h e characteristics and prognosis of newly diagnosed B-ALL (16,17).
Copy number variations (CNV; a.k.a. copy number aberrations [CNAs]) are a specific type of genetic abnormality with a high incidence in B-ALL (1,14,18), ranging from 1 Kb to less than 5 Mb (19). CNVs denote the deletion, insertion, replication, and multipoint variants of DNA fragments. Previously, the initial cognition of CNV was found in healthy people and correlated with neuropsychiatric disorders. Today, CNV is broadly recognized as a major cause of various solid tumors (20) and acute myeloid leukemia (21). This review primarily focuses on the CNV biomarker analysis in B-ALL and their prognostic significance.

CNV detection method
As CNVs are challenging to detect by karyotype analysis, fluorescence in situ hybridization (FISH), and PCR amplification; besides, their research and application are limited to some extent (22, 23). Indeed, FISH is traditionally used in CNV research but is limited to the imbalance design of both satisfying multi-genes location and the FISH gene-specific probes. With the advent of various sequencing technologies, array-based CNV analysis was commonly used for detecting genomic DNA fragments. For example, CNV can be recognized by array comparative genomic hybridization and singlenucleotide polymorphism arrays; however, the high cost and complex process of these techniques hinder their widespread use in clinical practice. In 2002, Schouten established multiplex ligation-dependent probe amplification (MLPA) assay to analyze the CNV spectrum; this technology is a fast and reliable gene CNV detection method that can detect the copy number changes of 45 gene probes simultaneously with high specificity and at a low cost (24). Kiss R et al. (25) proposed the digital MLPA-based approach based on the next-generation sequencing technology to detect hundreds of exon-positions CNV panels at the same time. The next-generation sequencing method can simultaneously detect sequence variation of a single base, insertion, or deletion of short fragments and CNV (19).
To date, many studies have investigated various software projects to examine copy number changes (26). Zhou B et al. (23) compared different sequencing depths (1×, 3×, and 5× coverages) using whole-genome sequencing by different sequencing libraries (short/3 kb/5 kb); they recommended that the gold standard for CNV detection was under the large library and low sequencing depth. Optical genome mapping is a new whole-genome sequencing method in which each DNA molecule is linearized and unfolded by nano-microfluidic CHIP with high-resolution fluorescence imaging (27,28).
All structural variations and CNVs can be detected by providing original DNA information for downstream applications of genomics. Unlike other traditional cytogenetic methods, optical genome mapping has a full coverage of all types of mutations, detects small tumor-related mutations, and has high consistency in detecting hematological malignanciesrelated chromosomal and DNA abnormalities. In addition, LüHMANN JL et al. (29) established that optical genome mapping was superior to any other traditional method in the area of detecting the classical gene deletions (e.g., IKZF1) and gene losses that were previously undetected (e.g., SETD2). Owing to the insensitivity of whole-genome sequencing hybridization and capture, the reads captured in an exon fragment vary markedly from sample to sample. Thus, new technologies emerged gradually, such as noninvasive prenatal testing technology, which could detect CNVs in tumor circulating free DNA of 7 MB size with >95% sensitivity and specificity (30).
Reportedly, RNA-seq is limited to detect CNVs in ALL as a result of mismatching B-allele frequency. BAŘINKA et.al (31). developed a robust tool RNAseqCNV package based on the normalized gene expression and minor allele frequency to classify arm-level CNVs. In addition, InferCNV was applied widely to identify large-scale chromosomal CNVs in tumor single-cell RNA sequencing (scRNA-seq) data. The basic idea is to compare the gene expression of each tumor cell with the average expression or "normal" reference cell gene expression in the whole genome to determine its expression intensity (32). However, the genomic location of specific CNVs is not available to precisely classify tumor and normal cells copy number spectrum. Considering the critical need for distinguishing normal cell types from malignant cells in the tumor microenvironment, copy number karyotype of tuments (CopyKAT), as an integrated Bayesian segmentation method, was developed to estimate the CNV spectrum, with an average genome resolution of 5 MB from the reading depth of highthroughput scRNA-seq data (33). Nevertheless, the research on CNV clones in relapsed B-ALL is limited. Despite being the preferred and widely used method for detecting CNVs in the related literature, MLPA might not be able to detect CNVs in samples presenting a low leukemia burden (carried <25% CNV clone). Moreover, CNVs in relapsed B-ALL remain unclear owing to limited paired B-ALL (newly diagnosed and relapsed) samples. The CNVs of relapsed B-ALL evolved from the diagnosis for examining specific gene content and clone size. By comparing the first-relapsed B-ALL to the newly diagnosed stage. RIBERA J et al. (48) established that CDKN2A/B, PAX5, and IKZF1 deletions were more frequent at relapse. Mullighan CG et al. (49) performed the genome-wide CNV and LOH analyses on matched diagnostic and relapse bone marrow samples from 61 pediatric patients with ALL, and identified a mean of 10.8 somatic CNV per B-ALL case and 7.1 CNVs per T-ALL case at diagnosis. In addition, they observed a significant increase in the mean number of CNVs per case in relapsed B-ALL samples (10.8 at diagnosis vs. 14.0 at relapse, P = 0.0005); however, no significant changes were observed in the lesion frequency in T-ALL. The majority (88.5%) of relapse samples harbored at least some of the CNAs present in the matched diagnosis sample, suggesting a common clonal origin, although 91.8% of samples showed a change in the pattern of CNVs from diagnosis to relapse. Of these cases, 34% acquired new CNVs, 12% exhibited loss of lesions present at diagnosis, and 46% both acquired new lesions and lost lesions present at diagnosis. Moreover, Ribera (48) compared CNVs at diagnosis and relapse, observing the trend to acquire homozygous CDKN2A/B deletions and a considerable increase in CNVs from diagnosis to the first relapse. Besides, evolution from an ancestral clone was the main pattern of clonal evolution. When focusing on the acquired CNVs in relapsed clones, gene alterations mostly correlated with proliferation and drug resistance.

IKZF1 gene deletions
The Ikaros Zinc Finger 1 (IKZF1) gene, located at 7p12.2, encodes 519 amino acids by 8 full-length exons (50). Exons are essential for Ikaros gene functions, except for exon 1 (which does not participate in transcription) and exons 2, 3, and 7 (undetermined significance). IKZF1 deletions in both coding and noncoding regions might interfere with the gene activity and promote B-ALL progression through specific targets. For example, EBF1, MSH2, and MCL1 genes, as the target genes of IKZF1, play a vital role in affecting B-cell differentiation (EBF1 gene), DNA repair (MSH2 gene), and anti-apoptosis (MCL1 gene). The primary functions of the IKZF1 gene include B-cell differentiation blocking, metabolic reprogramming, leukemia microenvironment adhesion, disease relapse, and drug resistance (51).
Increasing evidence indicated that IKZF1 deletions mediate cellular drug resistance and relapse. For example, Rogers et.al (52) established that the IKZF1 deletion was resistant to dexamethasone, asparaginase, and daunorubicin by upregulating the JAK/STAT pathway. In addition, the IKZF1 deletion affects sensitivity to cytarabine by downregulating the SAMHD1 pathway (52); STEEGHS et. al (14) suggested that the loss of IKZF1 caused prednisolone resistance by elevating intracellular ATP and glucose levels, whereas drug sensitivity was recovered by inhibition of glycolysis. Moreover, IKZF1 deletion events, accompanied by CREBBP deletion or mutation, were common in relapsed pediatric B-ALL patients, which could correlate with the selective pressure of chemotherapeutic drugs on tumor cells (8).
Notably, IKZF1 gene deletions comprise localized large fragment deletions, single exon deletions, and other nonlocalized deletions, among which localized large fragment deletions are the most common. The loss of IKZF1 can be separated depending on its functional effect. While IK1-IK3 is considered a functional subtype, other subtypes are dominantnegative isoforms (DN isoforms), that is, functional defect subtype. In addition, IK6, often located in the cytoplasm, is a functional defect subtype with the complete loss of N-terminal zinc finger structure due to exon 4-7 deletion. IK6 functions as DN effects by isolating normal cytoplasmic proteins (53). Lossof-function was designated as the total allelic inactivation. The loss of haploid dysfunction due to exon 2 deletion can decrease the Ikaros protein level.
Some studies reported IKZF1 deletions in around 15% of pediatric B-ALL cases and 30%-40% of adult B-ALL cases (40,54). Perhaps, IKZF1 deletions in pediatric B-ALL are a hallmark of high-risk stratification and relapse independently carried by 70% of high-risk pediatric B-ALL (45, 55, 56).
The prognostic impact of IKZF1 alterations in B-ALL remains debatable (58).
The response of early chemotherapy induction in patients with IKZF1 deletions was disappointing over the whole series. Several studies established that patients with IKZF1 lesions exhibited a high minimal residual disease (MRD) level (51,60,62). Reportedly, these patients could benefit more from intensive/alternate therapy than standard ones (4). Reportedly, the combination of vincristine and steroids in patients with IKZF1 deletions during maintenance treatment could be an effective and reasonable approach to prevent relapse. Dhedin N et al. (63) demonstrated that patients with IKZF1 deletions were likely to benefit from allogeneic HSCT (allo-HSCT) in terms of EFS (HR 0.42, 95% CI: 0.18-1.07, P = 0.025) and OS (HR 0.35, 95% CI: 0.16-0.75, P = 0.007), compared with non-IKZF1 alteration groups in adult Ph− B-ALL populations. However, whether the poor prognosis of IKZF1 overcame by stem cell transplantation warrants further investigation. CDKN2A/B deletion is the major suppressor gene CNV in chromosome 9p21 (66). Compared with children, the CDKN2A/B incidence rate is marginally higher in adults (P = 0.002) (67). Moreover, 24.6% (14/57) of Ph-like patients present with enriched biallelic loss of CDKN2A/B (68). Reportedly, this lesion was highly representative of high white blood cell count, older age at initial diagnosis, and often accompanied by IKZF1 deletions (called I&C) (36,69). Remarkably, clones with CDKN2A/B deletions detected in the initial diagnosis always persisted in relapse cases. Furthermore, CDKN2A/B presented a notable increase in the CNVs of relapse B-ALL (48).

CDKN2A/CDKN2B gene deletion
In some studies, pediatric B-ALL patients with CDKN2A/ B deletions exhibited a trend of shorter relapse time and EFS (35,67), although the OS rate remains debatable. Kathiravan et. al (35) indicated that the 28-month EFS of CDKN2A/B lesions in ICICLE (Indian adaption of UKMRC2007 protocol) was notably decreased (42% vs. 90%, P = 0.0004) compared with non- The frequency of adult CDKN2A/B deletions in the Ph-B-ALL group was much higher than in the Ph+ B-ALL group (39.7% vs. 24.7%, P = 0.041) (5). The prognostic value of CDKN2A/B in adults has been debated previously (35,41,44).

PAX5 gene deletion
The transcription factor paired box domain gene 5 (PAX5) was considered to regulate B-cell lineage differentiation and contribute to leukemogenesis in B-ALL (74,75). PAX5 acts on the downstream transcription factors E2A and EBF1 and is crucial for B-line differentiation (76). In PAX5-deficient mice, the development of B cells in the bone marrow was blocked in the early Pro-B stage (77). The alterations of PAX5 comprise partial exon deletion on chromosome 9 (14%) and amplification of exon 2 or 5, resulting in frameshift mutation (7%). PAX5 deletions might increase genetic instability. Consequently, the probability of a secondary strike markedly increases and induces the recurrence and development of leukemia. In a study, PAX5 deletions decreased leukemia cell viability by inducing apoptotic cell death using a new ribozyme-derived isotype-specific knockdown system in the B-ALL cell model (77). Furthermore, transplantation experiments and exhaustive sequencing validated that PAX5 deletion made it sensitive to malignant transformation by forming an abnormal progenitor cell population (78).
In pediatric B-ALL groups, the prognosis of PAX5 deletion was strongly dependent on IKZF1 codeletion (61,80). However, no significant prognostic correlation was observed in PAX5 deletions alone in children (74). In other words, the PAX5 -loss group presented no relapsing risk after excluding IKZF1 deletions. Indeed, double deletion of PAX5 and IKZF1 was improved by treatment intensification in MS2010, with 0% 5year CIR than 80.0% in MS2003 (P = 0.05).

Prognostic relevance of integrated CNV profiling
Extensive research integrated gene CNV profile into pediatric B-ALL risk stratification (17). Moorman AV et al. (34) identified an 8-gene CNV panel, including IKZF1, CDKN2A/B, PAR1, BTG1, EBF1, PAX5, ETV6, and RB1, for stratifying the pediatric B-ALL risk level known as the UKALL-CNV classifier ( Table 2). This tool has robust decision-making ability in intermediate-risk cytogenetics subgroups and even patients with different leukemia protocols baseline (37, 38). Besides, the UKALL-CNV classifier can refine the established cytogenetic risk groups.

Future perspectives
Many studies have proved that CNV is a common molecular abnormality in the development of B-ALL (48). Current evidence suggests that the CNV pattern of adult and pediatric B-ALL has a different cytogenetic abnormality and pathological significances. Moreover, growing evidence indicates that high number and diverse CNVs observed are acquired in the process of disease relapsing (37). This study mainly discussed the clinical significance of the CNV spectrum, which has been well recognized in patients with B-ALL. Among them, IKZF1, CDKN2A/B, and PAX5 are the leading prevalent gene alterations in B-ALL (47). Moreover, these CNVs in Ph-like and Ph+ B-ALL remain equally frequent (68). However, some research of gene prognostic value is inconsistent, which could be because of difference in enrolled patients and treatment regimen. Undoubtedly, CNVs guided the risk of relapsing and survival outcome of both pediatric and adult B-ALL (84). Intensive chemotherapy combined with allo-HSCT is expected to overcome the adverse impact of CNVs. Perhaps, the combination of intensive chemotherapy and allo-HSCT could overcome the adverse impact of CNVs.
Typically, the risk stratification of B-ALL based on the CNV profiles is largely limited to the pediatric population (36). Currently, the IKZF1 plus and UKALL-CNV classifier are broadly promoted in the adult B-ALL classification (5,37,43). Considering the different cytogenetic patterns of adults and children, the risk system in adults warrants revision in future. With the new exploration of new targets of rearrangement in B-ALL (e.g., DUX4, ZNF384, and MEF2D), the survival risk stratification system will be consistently updated in the future. Besides, further research will help identify new prognostic indicators and potential therapeutic targets. In conclusion, this review characterizes B-ALL-related copy number events, which is valuable for precise patient subgroup stratification. In addition, this study provides insights into the new immunotherapy-based approaches and tailored treatment strategies for patients with B-ALL. Nevertheless, additional multicenter survival data will be needed for further verification in the future.

Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.