*Review* **Advances in the Diagnosis and Treatment of Pediatric Acute Lymphoblastic Leukemia**

**Hiroto Inaba 1,2,\* and Ching-Hon Pui 1,2**


**Abstract:** The outcomes of pediatric acute lymphoblastic leukemia (ALL) have improved remarkably during the last five decades. Such improvements were made possible by the incorporation of new diagnostic technologies, the effective administration of conventional chemotherapeutic agents, and the provision of better supportive care. With the 5-year survival rates now exceeding 90% in highincome countries, the goal for the next decade is to improve survival further toward 100% and to minimize treatment-related adverse effects. Based on genome-wide analyses, especially RNAsequencing analyses, ALL can be classified into more than 20 B-lineage subtypes and more than 10 T-lineage subtypes with prognostic and therapeutic implications. Response to treatment is another critical prognostic factor, and detailed analysis of minimal residual disease can detect levels as low as one ALL cell among 1 million total cells. Such detailed analysis can facilitate the rational use of molecular targeted therapy and immunotherapy, which have emerged as new treatment strategies that can replace or reduce the use of conventional chemotherapy.

**Keywords:** acute lymphoblastic leukemia; pediatric; advances; diagnosis; treatment

#### **1. Introduction**

Approximately 6000 new cases of acute lymphoblastic leukemia (ALL) are diagnosed in the United States annually [1–4]. ALL is the most common pediatric cancer (representing approximately 25% of cancer diagnoses), and approximately 60% of all cases occur in children and adolescents younger than 20 years, with an annual incidence of 36.2 per 1 million persons and a peak age of incidence of two to five years (at which there are >90 cases per 1 million persons) [5]. ALL is diagnosed more frequently in boys than in girls, with a ratio of approximately 1.3:1. The annual incidence of ALL differs markedly according to race and ethnic group; there are 40.9 cases per million in the Hispanic population, 35.6 cases per million in the white population, and 14.8 cases per million in the black population [6]. ALL cases are broadly classified as B-ALL or T-ALL based on immunophenotyping, with B-ALL comprising approximately 85% of cases, although this percentage can differ depending on age at diagnosis, race, or ethnicity.

Currently, the survival of pediatric patients with ALL treated in high-income countries exceeds 90% (Figure 1) [1–4]. Chemotherapy is given in four important phases: remission induction, consolidation, reinduction (delayed intensification), and continuation (maintenance). Chemotherapy is administered based on stratified risk classification, as determined by clinical factors (e.g., age (1–9.9 years vs. <1 or ≥10 years) and white blood cell (WBC) counts (<50 × <sup>10</sup>9/L vs. ≥<sup>50</sup> × 109/L) at diagnosis), cytogenetic and genomic analysis of ALL cells, and response evaluation with a minimal residual disease (MRD) assay. Dosage adjustment based on pharmacodynamic and pharmacogenomic studies and supportive care (e.g., prevention and treatment of infection) have also contributed substantially to improved outcomes. Therefore, current dosages/schedules for "conventional" chemotherapy have been truly optimized.

**Citation:** Inaba, H.; Pui, C.-H. Advances in the Diagnosis and Treatment of Pediatric Acute Lymphoblastic Leukemia. *J. Clin. Med.* **2021**, *10*, 1926. https:// doi.org/10.3390/jcm10091926

Academic Editor: Tadeusz Robak

Received: 31 March 2021 Accepted: 25 April 2021 Published: 29 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Figure 1.** Overall survival of pediatric patients with acute lymphoblastic leukemia treated in the St. Jude Total Therapy studies.

With the current high rate of survival, further improvement in outcomes with conventional chemotherapy is challenging. In fact, there was very little improvement in 5-year overall survival (OS) between our two recent frontline ALL trials, St. Jude Total Therapy XV (5-year OS: 93.5%) and XVI (5-year OS: 94.3%) (Figure 1) [7,8]. Most of the conventional chemotherapy agents were approved by the US Food and Drug Administration before 1980 (Table 1), and their therapeutic intensity has been pushed to the limit of tolerance. Accordingly, further intensification of conventional chemotherapy could lead to only minimal improvement in overall outcomes while increasing adverse effects.

Recently, several molecular targeted agents and immunotherapy approaches have been introduced, and they promise to improve outcomes. For these agents to be used optimally, detailed genetic characterization of leukemia cells and response evaluation by MRD in individual patients are critical. In this review, we will review the genetic subgroups of ALL, the evaluation of MRD, and newer treatment strategies.


**Table 1.** Representative medications used in the treatment of patients with acute lymphoblastic leukemia and the year of US Food and Drug Administration approval.

**Table 1.** *Cont.*


\* Approval by the US Food and Drug Administration is not limited to indications for pediatric acute lymphoblastic leukemia. Abbreviation: NA, not approved.

#### **2. Genetic Characterization of Acute Lymphoblastic Leukemia**

The revolutionized approach to genomic analysis subdivides pediatric ALL into more than 30 genetic subgroups [9–11]. In B-ALL, recurrent genomic subtypes are characterized by chromosomal aneuploidy, i.e., hyperdiploidy (>50 chromosomes) or hypodiploidy (<44 chromosomes), and by rearrangements: *ETV6/RUNX1* fusion, *TCF3/PBX1* fusion, *BCR/ABL1* fusion, and *KMT2A* (*MLL*) rearrangement (Figure 2 and Table 2). Genetic abnormalities newly identified by comprehensive genomic analyses include *BCR/ABL1* like ALL (Ph-like ALL), intrachromosomal amplification of chromosome 21 (iAMP21), *DUX4*-rearranged ALL, *ZNF384*-rearranged ALL, *MEF2D*-rearranged ALL, *PAX5*-altered (PAX5alt) ALL, *NUTM1*-rearranged ALL, and *ETV6/RUNX1*-like ALL. Characterization of genetic abnormalities in ALL cells is important in order to identify unfavorable genetic abnormalities and to incorporate molecular targeted therapy to reduce the risk of relapse.


**Table 2.** Genetic subtypes and treatment approach.


**Table 2.** *Cont.*

\* Newly identified subgroups, necessary to confirm their prognosis in a larger number of patients. Abbreviations: MRD, minimal residual disease; iAMP21, intrachromosomal amplification of chromosome 21; ALL, acute lymphoblastic leukemia.

**Figure 2. Distribution of genetic subtypes** Genetic subgroups are listed based on the patients treated in St. Jude Total Therapy Study XVI and on patients with T-ALL who were treated in Children's Oncology Group studies and evaluated for genetics as part of the Therapeutically Applicable Research to Generate Effective Treatments initiative [11,12]. Percentages are the approximate incidence in pediatric ALL. B-ALL is categorized as low-, intermediate-, or high-risk disease. For T-ALL, no genetic subtypes are clearly associated with outcomes, but the group as a whole is considered an intermediate-risk group. Abbreviations: ALL, acute lymphoblastic leukemia.

#### **3. Low-Risk Genetic Subgroups**

#### *3.1. ETV6/RUNX1-Rearranged ALL*

*ETV6/RUNX1*-rearranged ALL represents approximately 20% of pediatric ALL and is associated with excellent outcomes [13]. Up to 5% of normal newborns carry the *ETV6/RUNX1* fusion at birth [14], and post-natal environmental or spontaneous oncogenic second hits are required to induce overt leukemia [15,16]. Patients with the *ETV6/RUNX1* fusion are good candidates for reductions in the intensity of chemotherapy if their initial MRD responses are good [17,18]. A randomized study of patients with standard-risk ALL enrolled on the Associazione Italiana di Ematologia e Oncologia Pediatrica–Berlin-Frankfurt-Münster (AIEOP–BFM) ALL 2000 protocol tested whether dose reductions by 30% for dexamethasone and by 50% for vincristine, doxorubicin, and cyclophosphamide during the delayed intensification phase resulted in outcomes comparable to those in the historical arm [19]. Although this study led to worse outcomes for the dose-reduction arm as a whole, outcomes in patients with *ETV6/RUNX1* fusion and in those aged 1 to 6 years were equivalent for the two arms. Furthermore, in the Tokyo Children's Cancer Study Group L92-13 study, which featured only 1 year of intensive chemotherapy, only two-thirds of the enrolled patients experienced continuous remission, but those with *ETV6/RUNX1* and *TCF3/PBX1* rearrangements had excellent outcomes with this abbreviated therapy [20]. Notably, patients with high hyperdiploidy fared poorly in this study.

#### *3.2. Hyperdiploid ALL*

Hyperdiploid ALL is the most common subtype of ALL, accounting for up to 25% of pediatric ALL. Different study groups have variously identified this subtype as having a DNA index of 1.16 or higher [21], a chromosome number of 51 to 67 [22], or trisomy of chromosomes 4 and 10 (double trisomy) [23]. Non-random gains of chromosome 4, 10, 14, 17, and 21 are common. Methotrexate is particularly useful for treating this subtype of ALL, and the disease response is influenced by the intracellular accumulation of active methotrexate polyglutamate metabolites (MTXPGs), which is higher in hyperdiploid ALL than in *ETV6/RUNX1* ALL, *TCF3/PBX1* ALL, or T-ALL [24–26]. This is partly due to the higher expression of the gene encoding the folate influx transporter *SLC19A1* in hyperdiploid ALL, resulting from the presence of a somatically acquired additional chromosome 21 on which this gene is located. Therefore, among patients with induction failure, those with hyperdiploid ALL had better outcomes than did those in other subgroups because they responded well to high-dose methotrexate, which is typically given as post-induction therapy, and these patients can be salvaged even without a hematopoietic cell transplant (HCT) [27].

Patients with *ETV6/RUNX1* fusion and hyperdiploidy and negative MRD on day 15 (as in St. Jude Total Therapy XVI) or day 19 (as in Total Therapy XV) and at the end of induction therapy have an excellent prognosis [11,17,18]. In St. Jude Total Therapy studies, patients with *ETV6/RUNX1* fusion and hyperdiploidy are provisionally treated in the low-risk (National Cancer Institute [NCI] standard-risk) arm regardless of their age or WBC count at diagnosis, but those patients with high MRD levels on day 15 (≥1%) or at the end of induction therapy (≥0.01%) or with extramedullary (central nervous system or testis) involvement are subsequently treated in the standard-risk (NCI high-risk) arm. This approach has been successful, with excellent outcomes for both subgroups [11,13,17].

#### *3.3. DUX4-Rearranged ALL*

*DUX4*-rearranged ALL is a newly identified subtype seen in 3% to 5% of pediatric ALL cases. The rearrangement occurs most commonly in the immunoglobulin heavy-chain locus (*IGH*) and results in the expression of DUX4 protein with a truncated C-terminus [28–30]. This truncated form binds an intragenic region of the ETS-family transcription factor *ERG* (ETS-related gene) and commonly results in the expression of a C-terminal ERG protein fragment that is a dominant-negative inhibitor of wild-type ERG function. *DUX4*-rearranged B-ALL has a unique immunophenotype (CD2 and CD371 positive), and a favorable outcome can be obtained, even with the deletion of *IKZF1*, by adjusting the intensity of the chemotherapy based on the MRD [31,32].

#### **4. High-Risk Genetic Subgroups in B-ALL**

#### *4.1. Hypodiploid ALL*

Hypodiploid ALL, which is defined by there being fewer than 44 chromosomes or a DNA index of less than 0.81, accounts for 1% to 2% of pediatric ALL. It is associated with poor outcomes, with reported EFS of 50% to 55% [33,34]. It can be classified into three distinct subtypes: near haploid (24 to 31 chromosomes), low hypodiploid (32 to 39 chromosomes), and high hypodiploid (40 to 43 chromosomes). Near-haploid ALL is associated with Ras pathway mutations (particularly in *NF1*) and *IKZF3* deletion [35]. Low-hypodiploid ALL is characterized by *TP53* mutations in the leukemia cells in more than 90% of cases and also in the germline in approximately 50% of patients, in addition to the somatic alterations in *IKZF2* and *RB1*. Therefore, patients with low-hypodiploid ALL should undergo germline testing for *TP53* germline pathogenic variants (i.e., Li–Fraumeni syndrome) to enable treatment modification to avoid the use of carcinogenic agents and for genetic consultation purposes [36]. It is important to distinguish "masked" hypodiploid ALL, in which the hypodiploid clone is duplicated, from true hyperdiploid ALL, considering the possible germline *TP53* mutations and the poor prognosis of hypodiploid ALL [37]. Recently, two multicenter studies demonstrated that HCT confers no benefit in hypodiploid

ALL, particularly for patients who are MRD negative after remission-induction therapy, for whom EFS was approximately 70% [33,34]. Therefore, patients with persistently positive MRD can be considered for treatment with molecular targeted agents such as BCL-2 inhibitors and PI3K inhibitors or with immunotherapy such as bispecific antibody therapy or chimeric antigen receptor (CAR) T-cell therapy [35,38,39].

#### *4.2. BCR/ABL1 (Philadelphia Chromosome)-Positive ALL*

*BCR/ABL1*-positive ALL accounts for approximately 2% to 3% of pediatric ALL [40]. Before tyrosine kinase inhibitors (TKIs) became available, the survival of patients who were treated only with conventional chemotherapy was dismal, and HCT from a matched related donor or an unrelated donor during the first remission provided a cure in only approximately 50% of children [41]. The combination of the TKI imatinib with multi-agent chemotherapy significantly improved outcomes, with 5-year disease-free survival increasing to 70% in the Children's Oncology Group AALL0031 study [42]. A second-generation TKI, dasatinib, targets both the ABL1 and SRC kinases, has activity against BCR/ABL1 that is approximately 300 times more potent than that of imatinib, and can cross the blood–brain barrier [40]. The Children's Oncology Group AALL0622 study, in which dasatinib was used at 60 mg/m2/day, showed no improvement in outcomes relative to those in the preceding AALL0031 study, in which imatinib (340 mg/m2/day) was given with the same chemotherapy backbone [43]. However, the Chinese Children's Cancer Group has shown that patients who received dasatinib (80 mg/m2/day) had significantly better EFS and OS and a lower relapse rate when compared with those who received imatinib (300 mg/m2/day) in a randomized study [44]. Ponatinib is one of a newer generation of TKIs and has potent activity in both wild-type *BCR/ABL1*-positive ALL and mutant forms (e.g., with the gatekeeper mutation ABL1 T315I) [40]. Treatment with ponatinib in combination with hyperfractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasone (hyper-CVAD), alternating with high-dose methotrexate and cytarabine, resulted in excellent 2-year EFS in adults with newly diagnosed *BCR/ABL1*-positive ALL [45]. Using ponatinib in combination with a pediatric regimen that includes asparaginase and glucocorticoids can be difficult as ponatinib is also associated with an increased risk of thrombosis and pancreatitis. In adult patients with *BCR/ABL1*-positive ALL, a chemotherapy-free regimen with glucocorticoid and dasatinib followed by blinatumomab and dasatinib was associated with a high molecular response and high survival rates with few adverse effects [46]. Nevertheless, the results of a recent preclinical study suggest that dasatinib may adversely affect the efficacy of blinatumomab [47]; additional studies are needed to determine whether these two agents should be used separately.

#### *4.3. BCR/ABL1 (Philadelphia Chromosome)-Like ALL*

*BCR/ABL1* (Philadelphia chromosome)-like ALL was initially identified as a subgroup of leukemias with a leukemic cell gene expression profile similar to that of *BCR/ABL1* positive ALL and frequent *IKZF1* alterations but without the *BCR/ABL1* fusion [48,49]. Although the percentage can vary depending on the ethnicity of the patients, this variant occurs in approximately 3% of pediatric ALL cases and is associated with worse outcomes. It is commonly seen in patients with NCI high-risk disease; however, *BCR/ABL1*-like ALL is also seen in patients with NCI standard-risk disease, and the outcome is associated with the MRD levels during and at the end of induction [50,51]. Many study groups have identified the genetic lesions associated with *BCR/ABL1*-like ALL, and these are classified in three main groups: JAK-STAT signaling activating mutations, *ABL1*-class fusions, and alterations that are less common and that involve other kinases [52,53].

JAK-STAT signaling activating mutations constitute the largest group and are genetically more diverse [52,53]. *CRLF2* rearrangements (*P2RY8/CRLF2* and *IGH/CRLF2*) and mutations (CRLF2 F232C) lead to CRLF2 overexpression, which may be detected by flow cytometry, and these mutations are present in approximately half of *BCR/ABL1*-like ALL cases, being more commonly seen in patients with Native American ancestry. Most of the

*JAK1* and *JAK2* mutations are seen in this group. Other JAK-STAT signaling activating mutations are present in approximately 10% of *BCR/ABL1*-like ALL cases and include *JAK2* fusions (translocations or interstitial deletions) that retain the tyrosine kinase domain, *EPOR* truncating rearrangements (e.g., with *IGH*, *IGK,* and *LAIR1*), *IL7R* insertion/deletions in the transmembrane domain, and deletions or mutations of *SH2B3* (a negative regulator of JAK-STAT signaling, the mutation of which augments JAK-STAT signaling). A JAK inhibitor, ruxolitinib, is currently being tested in clinical trials [53].

*ABL1*-class fusions involve *ABL1*, *ABL2*, *CSF1R*, *PDGFRB*, and, rarely, *PDGFRA* and *LYN* and are seen in 15% to 20% of *BCR/ABL1*-like ALL cases [52,53]. Pediatric patients with *ABL1*-class fusions have poor outcomes when treated with regimens that do not contain a TKI, even when they receive a high-risk chemotherapy regimen and/or HCT [54]. As seen in *JAK2* fusions, these are chimeric in-frame fusions that preserve the tyrosine kinase domain and are, therefore, sensitive to treatment with ABL1 inhibitors such as imatinib and dasatinib [52,55].

Other rare kinase-activating alterations include those in *NTRK3*, *FLT3*, *PTK2B*, and *TYK2*, and preclinical studies have shown the efficacy of treatment of these variants with a TRK inhibitor, an FLT3 inhibitor, an FAK inhibitor, and a TYK2 inhibitor, respectively [52,53].

#### *4.4. KMT2A-Rearranged ALL*

The *KMT2A* (*MLL*) gene is located on chromosome 11q23 and can be rearranged with more than 80 different partner genes, which are seen in both lymphoid and myeloid leukemia [56]. *KMT2A*-rearranged ALL is characterized by the CD10-negative pro-B cell phenotype with co-expression of myeloid markers. It accounts for approximately 5% of pediatric ALL and 75% of infant ALL. In infant ALL, *KMT2A* rearrangement is acquired in utero and is associated with dismal outcomes, especially in infants younger than 6 months at diagnosis with a presenting WBC count of ≥<sup>300</sup> × 109/L or with a poor prednisone response [56]. Although two international randomized studies were performed to examine standard vs. more intensive therapy before maintenance therapy (the Interfant-99 study) and myeloid- vs. lymphoid-type consolidation therapy (the Interfant-06 study), there were no significant differences in outcomes between interventions or studies [57,58]. *KMT2A* rearrangement results in the assembly of a unique multi-protein complex with DOT1L, BRD4, and menin [59]. Therefore, there is great potential for molecular targeted therapy with inhibitors of DOT1L, bromodomain, menin, and BCL-2. Immunotherapy with blinatumomab and autologous or allogeneic CAR T cells can be considered, although there is a possibility of a lineage switch to acute myeloid leukemia (AML) [56].

#### *4.5. MEF2D-Rearranged ALL*

*MEF2D*-rearranged ALL is seen in approximately 1% of pediatric ALL cases. The *MEF2D* gene can rearrange with several partner genes: *BCL9* (the most common partner), *CSF1R*, *DAZAP1*, *HNRNPUL1*, *SS18*, and *FOXJ2* [9,10]. *MEF2D*-rearranged ALL is characterized by older age at diagnosis (median, 14 years), mature B-cell leukemia morphology (large, densely basophilic, and heavily vacuolated leukemic blasts), a unique immunophenotype (weak or absent expression of CD10, high expression of CD38, and cytoplasmic immunoglobulin μ-chain), and poor outcome due to early relapse [60–62]. Exogenous expression of *MEF2D/BCL9* in a B-ALL cell line promoted cell growth, increased the expression of HDAC9 (a known *MEF2D* target), and induced resistance to dexamethasone [60]. Patient-derived leukemia cells were sensitive to histone deacetylase inhibitors (vorinostat and panobinostat) and to a proteasome inhibitor (bortezomib) in vitro and in xenograft models. *MEF2D/CSF1R* can be targeted by ABL1 inhibitors.

#### *4.6. TCF3/HLF-Rearranged ALL*

*TCF3/HLF*-rearranged ALL is a rare (representing <0.5% of cases) but very aggressive subtype of ALL. It is mostly resistant to conventional chemotherapy and has extremely poor outcomes even with intensified chemotherapy and HCT [63]. *TCF3/HLF*-rearranged

ALL is characterized by enrichment of stem cell and myeloid gene signatures, *PAX5* and *VPREB1* deletions, and Ras pathway gene mutations. *TCF3/HLF*-rearrangement plays a role as a pioneer transcription factor in recruiting EP300 to drive MYC, and EP300 inhibition reduces *TCF3/HLF*-dependent gene expression and ALL growth [64]. Drug activity profiling and preclinical studies have shown striking activity of a BCL-2 inhibitor, venetoclax [63]. Furthermore, all of nine patients with *TCF3/HLF*-rearranged ALL experienced molecular remission after being treated with blinatumomab, and four of them are in longterm remission after HCT, suggesting that an immunotherapy approach can overcome the chemotherapy resistance [65].

#### **5. Intermediate-Risk Genetic Subtypes in B-ALL**

#### *5.1. TCF3/PBX1-Rearranged ALL*

*TCF3/PBX1*-rearranged ALL is generated with the t(1;19)(q23;p13) translocation and is present in approximately 2% to 5% of pediatric ALL cases, commonly expressing cytoplasmic μ chain (a pre-B phenotype) [66]. As with *ETV6/RUNX1*-rearranged ALL, the preleukemic *TCF/PBX1* gene fusion is present in approximately 0.6% of healthy newborns [67]. The incidence of this leukemia variant is higher in African Americans [68], and a genome-wide association study identified a germline risk locus in an intergenic region between *BCL11A* and *PAPOLG*: rs2665658 [69]. In the St. Jude Total XV study, which eliminated cranial irradiation, *TCF/PBX1*-rearranged ALL was associated with a higher incidence of CNS relapse but a lower incidence of hematologic relapse compared to other forms of B-ALL [7,66]. In patients treated in the Total XVI study, the incidence of CNS relapse was reduced as a result of the increased frequency of early intrathecal treatments [8]. In the TCCSG L92-13 study, *TCF3/PBX1*-rearranged ALL had excellent outcomes with 1 year of intensive chemotherapy from diagnosis [20].

#### *5.2. Intrachromosomal Amplification of Chromosome 21 (iAMP21)*

Intrachromosomal amplification of chromosome 21 (iAMP21) ALL is characterized by the presence of additional copies of a region of chromosome 21 that includes *RUNX1* (five or more copies per cell), and it can be associated with the germline Robertsonian translocation rob(15;21) [70,71]. iAMP21 ALL is seen in approximately 1% to 2% of pediatric ALL cases and is associated with older age (median, 9 years) and low WBC counts. Secondary cytogenetic and genetic changes include the gain of chromosome X, the loss or deletion of chromosome 7, *ETV6* and *RB1* deletions, and *SH2B2* inactivation through copy numberneutral loss of heterozygosity of chromosome 12q [72,73]. Patients with iAMP21 had dismal outcomes when treated with a low-intensity NCI standard-risk regimen [74,75]. Although intensified treatment has significantly improved the outcomes for these patients, their EFS remains inadequate at approximately 70%. Therefore, they can also be considered candidates for recently introduced novel therapies.

#### *5.3. PAX5-Driven Subtypes: PAX5alt and PAX5 p.Pro80Arg*

PAX5 is the B-lymphoid transcription factor that is essential for early stages of Bcell development [76,77]. Germline alterations of the *PAX5* gene predispose patients to ALL, and somatic alterations of *PAX5* are commonly seen in pediatric ALL (e.g., *PAX5* focal deletions are present in approximately 30% of *ETV6/RUNX1*-rearranged ALL) [77]. The two distinct disease-initiating alterations of *PAX5* that result in PAX5alt and PAX5 p.Pro80Arg ALL account for approximately 3% to 5% and less than 1% of childhood ALL, respectively [9,10]. PAX5alt B-ALL is characterized by diverse *PAX5* alterations, including rearrangements (most commonly with *ETV6* or *NOL4L*), sequence mutations, and intragenic amplification. PAX5 p.Pro80Arg is characterized by universal p.Pro80Arg mutation with deletion or mutation of the remaining allele and alterations in Ras and JAK-STAT pathway genes. Patients with PAX5alt or PAX5 p.Pro80Arg B-ALL have an intermediate prognosis [9,10].

#### *5.4. ZNF384-Rearranged ALL*

*ZNF384*-rearrangement is seen in approximately 1% to 2% of childhood ALL cases and in half of B/myeloid mixed-phenotype acute leukemia (MPAL) cases in children. This rearrangement has more than 10 partner genes, such as *EP300*, *TCF3*, *TAF15*, and *CREBBP* [62,78,79]. In B-ALL, the age of onset and the prognosis differ according to the fusion partner: with the *EP300/ZNF384* fusion, the median age of onset is 11 years and outcomes are excellent, whereas with the *TCF3/ZNF384* fusion, the median age of onset is 5 years and there are occasional late relapses [78,80]. The immunophenotype of *ZNF384*-rearranged B-ALL is characterized by negative or weak expression of CD10 and aberrant expression of CD13 and/or CD33 [78,80]. As with *ETV6/RUNX1*-rearranged and *TCF3/PBX1*-rearranged B-ALL, a study in monozygotic twins showed that *TCF3/ZNF384* fusion can occur in utero, suggesting that a fetal hematopoietic progenitor is the cell of origin in this ALL subgroup [81]. Importantly, the secondary genomic alterations and gene expression profiles for *ZNF384*-rearranged B-ALL and B/myeloid MPAL cases are essentially indistinguishable, which suggests that ALL-directed therapy should be initiated for patients with newly diagnosed B/myeloid MPAL [79]. Due to its inherent lineage plasticity, *ZNF384*-rearranged leukemia may develop a lineage switch at relapse (from ALL to AML or vice versa) under the selective pressure of conventional chemotherapy or immunotherapy.

#### **6. Other Newly Identified B-ALL Subtypes**

#### *6.1. ETV6/RUNX1-Like ALL*

*ETV6/RUNX1*-like ALL is seen in 1% to 3% of pediatric ALL cases and is particularly common in younger children [9,10,30]. It has a similar gene expression profile and immunophenotype to *ETV6/RUNX1*-rearranged ALL but lacks the *ETV6/RUNX1* fusion. Within this group, alterations in *ETV6, IKZF1*, and *TCF3* have been reported. As the number of patients identified to date is small and several relapses have been reported, it is important to evaluate the actual outcomes of patients in this group, which appear to be worse than those of patients with *ETV6/RUNX1*-rearranged ALL.

#### *6.2. NUTM1-Rearranged ALL*

*NUTM1*-rearranged ALL is seen in 5% to 7% of all infants with ALL and represents 21.7% of non-*KMT2A*-rearranged infant ALL, but it is very rare in children (accounting for less than 1% in that population) [9,10,82,83]. Partner genes include *ACIN1*, *CUX1*, *BRD9*, and *ZNF618*. In an international study, the 4-year OS in 45 infants and 36 children was 100%, which is indicative of a favorable genetic subtype, although further studies are required to confirm this finding and to determine whether a reduction in treatment intensity is possible [82].

#### **7. T-Acute Lymphoblastic Leukemia**

T-ALL represents approximately 12% to 15% of pediatric ALL and is characterized by having an incidence in boys that is two to three times that in girls; a higher proportion of patients with African ancestry, in whom the rate is twice that in patients of European ancestry; high initial WBC counts; and higher frequencies of mediastinal mass and CNS involvement [12,84]. The higher incidence in boys can be partly explained by inactivating mutations or deletions of the tumor suppressor gene *PHF6* on chromosome X, which are seen in 16% of pediatric T-ALL cases [85]. The genetic alterations in T-ALL are diverse, and no clear associations with outcomes have yet been identified. Hence, unlike B-ALL, T-ALL lacks a consensus genetic classification with prognostic implications. In most cases of T-ALL, there is aberrant expression of transcription factors and oncogenes, including *TAL1*, *TAL2*, *LYL1*, *LMO1*, *LMO2*, *TLX1* (*HOX11*), *TLX3* (*HOX11L2*), and *HOXA* [86]. *NOTCH1* activating mutations and alterations in *CDKN2A/CDKN2B* are seen in more than 70% of cases, and *MLLT10* and *KMT2A* rearrangements are each seen in 5% of cases. Approximately 25% of patients have JAK-STAT activating mutations, and *ABL1* fusions

with *BCR* and *NUP214* are occasionally detected [86]. These patients are candidates for treatment with JAK inhibitors and ABL1 inhibitors, respectively.

In most studies, the survival of patients with T-ALL is 5% to 10% worse than that of patients with B-ALL [12]. With regard to conventional chemotherapy, the treatment component of the BFM IB phase that includes cyclophosphamide, cytarabine, and mercaptopurine is of greater importance for T-ALL than for B-ALL [87]. In one study, patients with T-ALL who received nelarabine had significantly fewer incidences of CNS relapse (isolated and combined) when compared to patients who did not receive nelarabine [88]. However, approximately 90% of the total patients and all of the nelarabine-treated patients received cranial irradiation in this randomized study; therefore, the efficacy of nelarabine should be confirmed in patients whose disease is managed with intrathecal therapy only. The results of the recent randomized study of bortezomib are described below [89].

#### *Early T-Cell Precursor ALL*

Early T-cell precursor (ETP) ALL accounts for 10% to 15% of T-ALL, having a specific immunophenotype of early T-cell development (cytoplasmic CD3+, CD5weak, CD8−, CD1a−) with aberrant expression of myeloid and/or early progenitor cell markers [90]. The genetic features of this subtype are similar to those of hematopoietic stem cells; it is characterized by alterations in transcriptional regulators, epigenetic regulation, and JAK-STAT and Ras pathway genes [86,91]. Furthermore, ETP ALL shares genomic features with T/myeloid MPAL, with frequent biallelic *WT1* alterations and signaling pathway mutations (e.g., in the JAK-STAT and *FLT3* pathways) [79]. ETP-ALL is usually glucocorticoid resistant, has a higher incidence of induction failure, especially after the BFM IA phase [92,93], and is historically associated with worse outcomes [90,94]. However, ETP-ALL responds to a regimen that includes cyclophosphamide, cytarabine, and mercaptopurine (e.g., the BFM IB phase), and its outcomes are approaching those of non-ETP T-ALL [92,93,95]. The results of a preliminary study suggested that patients with ETP-ALL would benefit from treatment with venetoclax, a BCL-2 inhibitor [96].

#### **8. Minimal Residual Disease**

Although genetic subclassification is essential for risk stratification, MRD has equally important prognostic and therapeutic impact [97–99]. MRD has been quantified by multiparametric flow cytometry or by allele-specific oligonucleotide PCR analysis. The flow cytometric assay uses the leukemia-specific aberrant immunophenotype, has a typical sensitivity of 0.01%, and can be applied to almost all cases of ALL [98,99]. It is rapid, enables accurate quantification of ALL cells, and provides an overview of the hematopoietic cell population status. However, it can be difficult to achieve sensitivity better than 0.01%, and the assay may fail to detect an ALL population that has undergone a phenotypic change, especially after immunotherapy targeting CD19 and/or CD22. The PCR assay amplifies leukemia-specific fusion transcripts (available for approximately 40% of ALL cases) or immunoglobulin (Ig) or T-cell receptor (TCR) genes (available for approximately 90% of ALL cases) with a sensitivity of 0.001%, 10 times that of the flow cytometry assay [98,99]. In RT-PCR analysis of fusion transcripts, there is a possibility of RNA degradation or cross-contamination from other samples. For Ig and TCR DNA, tailor-made primers are needed for each patient. Furthermore, ALL can be oligoclonal and may escape detection by clonal evolution during treatment. Recently, next-generation sequencing (NGS) of Ig or TCR genes has been applied for MRD detection (NGS MRD) with sensitivity as low as 0.0001% (equivalent to detecting one ALL cell among 1 million total cells) [100,101]. The use of universal primers enables the detection of clonal evolution and can also detect the background repertoire of normal B and T cells. With this technology, negative NGS MRD at the end of induction has been associated with 100% OS among NCI standardrisk patients [102]. In pediatric patients with ALL who received HCT, negative pre-HCT MRD and post-HCT MRD were associated with significantly fewer relapses and better survival [103]. The NGS MRD assay might not be affected by phenotypic changes after

immunotherapy, and negative NGS MRD after CAR T-cell therapy was also associated with better outcomes as compared with those in patients with positive NGS MRD among the patients with negative flow MRD [104]. These clinical benefits will result in expanded use of NGS MRD in contemporary protocols.

When considering risk stratification, clinicians should consider MRD levels in combination with genetic classification and clinical factors (e.g., age, WBC counts at diagnosis, and lineage) [17,18,97,105]. Patients with favorable genetic features clear MRD faster than do those with unfavorable genetics or T-ALL. Furthermore, as seen in *ETV6/RUNX1* rearranged and hyperdiploid ALL, some patients with favorable genetics but slow MRD clearance can be cured by intensifying their post-remission chemotherapy [11,17,27]. Conversely, patients with high-risk genetics have inferior outcomes even when they have undetectable MRD at the end of induction therapy [11,17,18]. It is also important to evaluate whether more sensitive NGS MRD can identify patients with better outcomes among those patients with high-risk genetic features. Furthermore, patients with T-ALL who had negative MRD (<103) on day 78 had a cumulative risk of relapse similar to that of patients who had negative MRD on day 33 [87]. In such patients, the MRD level on day 33 was not relevant, suggesting that the MRD response to the BFM IB phase (two courses of cyclophosphamide, cytarabine, and mercaptopurine) is critical in T-ALL.

#### **9. Emerging Therapy: Molecular Targeted Therapy**

#### *9.1. Tyrosine Kinase Inhibitors*

Tyrosine kinase inhibitors have been employed in combination with standard chemotherapy to improve its efficacy (Table 1). As described earlier, ABL1 inhibitors (e.g., imatinib, dasatinib, nilotinib, and ponatinib) are used to treat patients with *BCR/ABL*positive ALL and *ABL1*-class fusions that occasionally occur in *BCR/ABL*-like ALL and T-ALL [40,53,55]. Ruxolitinib is being tested in clinical trials for patients with JAK-STAT activating mutations as seen in *BCR/ABL*-like ALL and T-ALL (including ETP-ALL) [53]. Currently, however, this targeted approach is limited to less than 10% of pediatric ALL cases. Further identification of ALL driving mutations and their targets will expand the use of TKIs. In this regard, ex vivo leukemia drug-sensitivity profiling identified that 44.4% of childhood T-ALL samples and 16.7% of adult T-ALL samples as being sensitive to dasatinib through the inhibition of preTCR-LCK signaling [106].

#### *9.2. BCL-2 and BCL-XL Inhibitors*

Members of the B-cell lymphoma 2 (BCL-2) protein family play critical roles in the intrinsic mitochondrial apoptosis pathway through interactions between pro- and antiapoptotic proteins (Table 1) [107]. Venetoclax is a selective inhibitor of BCL-2 and displaces the pro-apoptotic proteins BIM and BAX, which leads to mitochondrial outer membrane permeabilization, cytochrome c release, and the activation of intracellular caspases, resulting in apoptosis. Preclinical studies have shown that venetoclax is active for leukemias in the high-risk genetic group, such as *KMT2A*-rearranged ALL [108], hypodiploid ALL [38], *BCR/ABL*-positive ALL [109], *TCF3/HLF*-rearranged ALL [63], and T-ALL (including ETP-ALL) [110,111]. Low expression of *CELSR2* is associated with the overexpression of *BCL2* and glucocorticoid resistance in ALL cells [112]. Venetoclax mitigated glucocorticoid resistance and had synergistic effects with prednisolone and dexamethasone.

Phase I studies of venetoclax in combination with chemotherapy in pediatric and young adult patients with ALL have shown the regimen to be well tolerated with preliminary efficacies [113]. As the results of a preclinical study suggested that ALL cells were dependent on both BCL-2 and BCL-XL, navitoclax (a BCL-2 and BCL-XL inhibitor) was tested in combination with venetoclax and chemotherapy for pediatric and adult patients with relapsed/refractory ALL or lymphoblastic lymphoma [114]. Among 47 heavily pre-treated patients, the complete remission rate was 60%, showing the regimen to have promising efficacy.

#### *9.3. Proteasome Inhibitors*

Proteasome inhibitors have shown efficacy in ALL and work synergistically with chemotherapy agents such as corticosteroids and doxorubicin (Table 1) [115]. In 22 children with relapsed ALL treated with bortezomib in combination with vincristine, dexamethasone, pegaspargase, and doxorubicin, the overall response rate was 73% [116]. In a randomized study of patients with newly diagnosed T-ALL or T-lymphoblastic lymphoma (T-LLy), adding bortezomib to the induction and delayed intensification phases was associated with better outcomes, as compared to those in patients who did not receive bortezomib, in patients with standard-risk and intermediate-risk T-ALL, as well as in those with T-LLy [89]. However, addition of bortezomib was associated with worse outcomes in patients with high-risk T-ALL. Newer proteasome inhibitors (carfilzomib and ixazomib) are under investigation.

#### *9.4. Other Molecular Targeted Therapies*

Dysregulation of the PI3K/AKT/mTOR pathway is frequently observed in ALL and is associated with resistance to chemotherapy [117,118]. mTOR inhibitors have been shown to inhibit ALL growth and reverse glucocorticoid resistance and to work synergistically with other chemotherapeutic agents, such as dexamethasone, vincristine, and doxorubicin (Table 1) [119–121]. A phase I study of everolimus with vincristine, prednisone, pegasparagase, and doxorubicin in children and adolescents with ALL in first marrow relapse occurring more than 18 months after first complete remission showed that the regimen was tolerable [122]. Nineteen (86%) of 22 enrolled patients had a second complete remission, and 13 (68%) of them had negative MRD.

Epigenetic modification, the biochemical alteration of chromatin, has been implicated in the pathogenesis of cancer [123]. Instead of changes in the nucleotide sequence, epigenetic modifications involve DNA methylation and histone modification, which affect the activity of genes and their cellular expression. These modifications can silence tumor suppressor genes or activate oncogenes. They are prevalent in ALL and are associated with chemotherapy resistance and relapse [124]. Epigenetic modifications may be reversible with targeted agents such as DNA methyltransferase inhibitors and histone deacetylase inhibitors (Table 1). In a phase 1 study of decitabine and vorinostat in combination with vincristine, dexamethasone, mitoxantrone, and pegaspargase, 22 children and adolescents with relapsed or refractory ALL were treated [125]. Although this regimen was associated with a high incidence of infectious complications, nine patients (39%) had a complete response, and potent pharmacodynamic modulations of biological pathways associated with antileukemic effects were observed.

#### **10. Emerging Therapy: Immunotherapy**

Three major categories of immunotherapy are currently in use for pediatric ALL (Figure 3 and Table 1): bispecific antibodies (e.g., blinatumomab), CAR T cells, and antibody–drug conjugates (e.g., inotuzumab) [126]. Immunotherapy has been used mostly for B-ALL because the surface markers CD19, CD20, and CD22 are expressed only on B cells and not on hematopoietic stem cells or other tissues. Such therapy can eradicate not only B-ALL but also normal B cells, thereby causing hypogammaglobulinemia, which can be managed by intravenous or subcutaneous immunoglobulin administration. For T-ALL, antibody therapy (e.g., with daratumumab against CD38) and CAR T cells (e.g., anti-CD1a, CD5, and CD7) are under investigation (Figure 3).

**Figure 3. Immunotherapy in acute lymphoblastic leukemia.** Abbreviations: ALL, acute lymphoblastic leukemia; CAR, chimeric antigen receptor; TSLPR, thymic stromal lymphopoietin receptor.

#### *10.1. Bispecific Antibody Therapy*

Blinatumomab has bispecific single-chain Fv fragments that link CD3+ T cells to CD19+ leukemia cells and cause a cytotoxic immune response (Figure 3 and Table 1) [127,128]. It is approved for use in pediatric and adult relapsed/refractory and MRD-positive B-ALL by the US Food and Drug Administration. The main adverse effects are cytokine release syndrome and neurotoxicity, which coincide with T cell activation. Two randomized studies in children, adolescents, and young adults with intermediate-risk or high-risk relapsed/refractory B-ALL showed blinatumomab to have benefits over intensive consolidation chemotherapy [129,130]. The loss of CD19 expression is a major mechanism of resistance to blinatumomab treatment and is also observed with CAR T cell therapy. Acquired genetic mutations in *CD19* exons 2–5 or alternative splicing at exon 2 produce a truncated protein with a nonfunctional or absent transmembrane domain and/or no antibody binding site [131,132]. Sustained CD19-antibody pressure can result in lineage switches as described in *KMT2A*- and *ZNF384*-rearranged B-ALL [133,134]. An alteration in CD81, which is a chaperone protein for the maturation and trafficking of the CD19 molecule from the Golgi apparatus to the cell surface, has been also reported [135].

#### *10.2. Chimeric Antigen Receptor (CAR) T Cells*

CAR T cells express single-chain Fv fragments against B-lineage markers (e.g., CD19, CD22, or both) with intracellular signaling domains such as 4-1BB or CD28 with CD3ζ [136]. A phase 2 international study of anti-CD19 CAR T cells (tisagenlecleucel) in pediatric and young adult patients with relapsed/refractory B-ALL showed a complete remission rate of 81% at 3 months and EFS and OS of 73% and 90%, respectively, at 6 months [137]. Currently, tisagenlecleucel is approved for patients up to 25 years of age with B-ALL that is refractory or in a second or later relapse. Several groups consider CAR T cells to be curative therapy, although others view them as a bridging therapy to HCT. As with blinatumomab, cytokine release syndrome and neurotoxicity are commonly seen with CAR T-cell therapy [138]. Preemptive administration of tocilizumab (an anti-IL-6 receptor antibody) decrease the incidence of severe cytokine release syndrome without compromising the efficacy of CAR T cells [139]. CAR T-cell recipients are also at high risk for infection, and they should be considered for bacterial and fungal prophylaxis until their neutropenia resolves, in addition to immunoglobulin supplement and *Pneumocystis jirovecii* pneumonia prophylaxis [140].

Mechanisms of resistance to CAR T-cell therapy include the loss of CAR T-cell persistence and B-cell aplasia and antigen loss on ALL cells [141,142]. In the former scenario, the type of co-stimulatory molecule (e.g., 4-1BB vs. CD28), rejection due to the murine component in tisagenlecleucel, and T-cell exhaustion are considered important factors. The use of two co-stimulatory molecules or new types of co-stimulatory molecule; humanized CAR T cells; in vivo stimulation with a CD19 vaccine, cytokines, or check point inhibitors; or early collection of T cells during treatment for high-risk patients may overcome this issue. With regard to target antigen loss, CAR T cells that can target other antigens (e.g., CD22 or the thymic stromal lymphopoietin receptor) or that can simultaneously target dual antigens (e.g., CD19/CD22) and the administration of two independent CAR T cells that target different antigens are being investigated [143–147].

For extramedullary relapse (e.g., in the CNS and testes), CAR T cells can migrate and show anti-leukemia effects; therefore, they can be considered not only for isolated bone marrow relapses but also for isolated or combined extramedullary relapses, thereby avoiding radiation therapy [148,149].

#### *10.3. Antibody-Drug Conjugates*

Inotuzumab ozogamicin is an anti-CD22 antibody that is linked to calicheamicin, a cytotoxic antitumor antibiotic that causes double-strand DNA breaks [150]. Inotuzumab is currently approved for use in adult patients with relapsed/refractory B-ALL. It is associated with sinusoidal obstruction syndrome, especially after HCT [150]. Fractionated weekly dosing of inotuzumab at the dose lower than a single dose given every 3–4 weeks and a longer interval between inotuzumab administration and HCT (i.e., 2 months or more) can reduce the incidence of this syndrome [151]. Additionally, it is recommended to use prophylactic pharmacologic agents (e.g., ursodiol), to limit the inotuzumab use to two cycles if HCT is planned, and to avoid HCT conditioning regimens that contain dual alkylating agents (e.g., thiotepa and melphalan) and concomitant hepatotoxic drugs (e.g., azoles) [152]. In a pediatric phase I study that used fractionated weekly dosing for relapsed/refractory B-ALL, complete remission was seen in 80% of the patients and 84% of those with available flow cytometry data had negative MRD [153].

#### **11. Conclusions**

The diagnosis of ALL, the treatment of patients, and the evaluation of the treatment response have undergone remarkable improvement. The detailed genetic characterization of ALL cells, functional genomics and proteomics, and drug sensitivity assays with ex vivo and patient-derived xenograft (PDX) models for molecular targeted agents and immunotherapy will lead to new therapeutic strategies. Furthermore, the evaluation of germline genetics can lead to an understanding of leukemogenesis, cancer predisposition, and the differences in drug response and metabolism (pharmacogenomics). Basic, translational, and clinical research on ALL will not end until all patients can be cured without acute complications or late sequelae.

**Author Contributions:** Conceptualization, H.I. and C.-H.P.; data curation, H.I.; writing—original draft preparation, H.I.; writing—review and editing, H.I. and C.-H.P.; funding acquisition, H.I. and C.-H.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** Supported in part by Cancer Center Core Grant CA21765 from the National Institutes of Health and by the American Lebanese Syrian Associated Charities (ALSAC). The funding organizations had no role in the design and conduct of the review; the collection, management, analysis, and interpretation of the data; the preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data sharing not applicable.

**Acknowledgments:** The authors thank Keith A. Laycock for scientific editing of the manuscript.

**Conflicts of Interest:** The authors have no conflict of interest, including specific financial interests, relationships, or affiliations relevant to the subject of this manuscript.

#### **References**


**Ilaria Iacobucci 1,\*, Shunsuke Kimura <sup>1</sup> and Charles G. Mullighan 1,2,\***


**Abstract:** Acute lymphoblastic leukemia (ALL) is the most successful paradigm of how risk-adapted therapy and detailed understanding of the genetic alterations driving leukemogenesis and therapeutic response may dramatically improve treatment outcomes, with cure rates now exceeding 90% in children. However, ALL still represents a leading cause of cancer-related death in the young, and the outcome for older adolescents and young adults with ALL remains poor. In the past decade, next generation sequencing has enabled critical advances in our understanding of leukemogenesis. These include the identification of risk-associated ALL subtypes (e.g., those with rearrangements of *MEF2D*, *DUX4*, *NUTM1*, *ZNF384* and *BCL11B*; the PAX5 P80R and IKZF1 N159Y mutations; and genomic phenocopies such as Ph-like ALL) and the genomic basis of disease evolution. These advances have been complemented by the development of novel therapeutic approaches, including those that are of mutation-specific, such as tyrosine kinase inhibitors, and those that are mutationagnostic, including antibody and cellular immunotherapies, and protein degradation strategies such as proteolysis-targeting chimeras. Herein, we review the genetic taxonomy of ALL with a focus on clinical implications and the implementation of genomic diagnostic approaches.

**Keywords:** B-ALL; *DUX4*; *IKZF1*; *PAX5*; Ph-like; *ZNF384*; *NUTM1*; T-ALL; *NOTCH1*; *BCL11B*; transcriptome; genome

#### **1. Introduction**

Acute lymphoblastic leukemia (ALL) is the most frequent childhood tumor and despite cure rates now exceeding 90% in children, outcomes for older children and adults remain poor with cure rates below 40% in those over the age of 40 [1–3], despite pediatricinspired chemotherapy regimens [4]. This discrepancy is in part attributable to the different prevalence of genetic alterations across age. ALL may be of B- (B-ALL) or T-lymphoid (T-ALL) lineage, and comprises over thirty distinct subtypes characterized by germline and somatic genetic alterations that converge on distinct gene expression profiles [5–12]. These subtypes are defined by disease-initiating recurrent chromosomal gains and losses (hyperand hypodiploidy, and complex intrachromosomal amplification of chromosome 21); chromosomal rearrangements that deregulate oncogenes or encode chimeric fusion oncoproteins, importantly often including cryptic rearrangements not identifiable by conventional cytogenetic approaches, such as *DUX4* and *EPOR* rearrangements; subtypes defined by single point mutations (e.g., PAX5 P80R or IKZF1 N159Y); subtypes defined by enhancer hijacking (e.g., *BCL11B*-rearrangements in T-ALL and lineage ambiguous leukemia) [5]; and subtypes that "phenocopy" established subtypes, with similar gene expression profile but different founding alterations (e.g., *BCR-ABL1*-like ALL and *ETV6-RUNX1*-like ALL) [7,13–15]. Secondary somatic DNA copy number alterations and sequence mutations are also important in leukemogenesis and treatment response, and their nature and

31

**Citation:** Iacobucci, I.; Kimura, S.; Mullighan, C.G. Biologic and Therapeutic Implications of Genomic Alterations in Acute Lymphoblastic Leukemia. *J. Clin. Med.* **2021**, *10*, 3792. https://doi.org/10.3390/jcm 10173792

Academic Editors: Rupert Handgretinger and Tadeusz Robak

Received: 22 July 2021 Accepted: 18 August 2021 Published: 25 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

prevalence vary according to the ALL subtype [6]. Multiple genes are associated with predisposition to ALL, including polymorphic variants in *ARID5B, BAK1, CDKN2A*, *CDKN2B*, *CEBPE*, *ELK3*, *ERG*, *GATA3*, *IGF2BP1*, *IKZF1*, *IKZF3*, *LHPP*, *MYC*, *PTPRJ*, *TP63* and the *BMI1-PIP4K2A* locus or rare mutations in *PAX5*, *TP53*, *IKZF1* and *ETV6* [16]. Several are associated with ALL subtype, for example, variants in *GATA3* have been associated with an increased risk of Philadelphia- like (Ph-like) ALL in patients of Hispanic ancestry [17], variants in *TP63* and *PTPRJ* with *ETV6-RUNX1* ALL [18] and in *ERG* with *TCF3-PBX1* ALL and African American ancestry [19,20]. A variant in the deubiquitinase gene *USP7* has been instead associated with risk of T-lineage ALL [19].

Accurate identification of the genetic abnormalities that drive ALL is important to risk stratify disease, and to guide the incorporation of molecular targeted therapeutic approaches to reduce the risk of relapse. This has been previously relied upon conventional karyotyping, fluorescence in situ hybridization (FISH) and targeted-molecular analyses. However, studies from this past decade have highlighted the importance of next generation sequencing (NGS) approaches to identify cryptic genetic rearrangements, structural DNA variation and gene expression signatures otherwise not identifiable that demand a revision of diagnostic approaches. This review describes the current genomic landscape of B- and T-ALL, highlighting their genetic characterization and diagnostic classification, clinical features, and therapeutic implications.

#### **2. B-Cell Precursor Acute Lymphoblastic Leukemia**

#### *2.1. Previously Established Subtypes with Recurring Chromosomal Abnormalities*

Prior the advent of NGS, classification of ALL has been relied on conventional karyotyping, FISH and targeted-molecular analyses for the identification of recurring chromosomal abnormalities including aneuploidy, chromosomal rearrangements and/or known gene fusions (Figure 1).

#### 2.1.1. Subtypes with Chromosomal Aneuploidy

Chromosomal aneuploidies [21], such as hyperdiploidy and hypodiploidy, are generally early initiating events acquired prenatally during fetal hematopoiesis and likely require secondary cooperating oncogenic insults to promote leukemia development [22].

High hyperdiploidy (modal number of 51–67 chromosomes, with nonrandom gains most commonly of chromosomes X, 4, 6, 10, 14, 17, 18, and 21) is present in 25–30% of ALL in children and is associated with young age (3–5 years) at diagnosis and favorable outcome [23,24]. Mutations of genes encoding mediators of Ras signaling (*KRAS*, *NRAS*, *FLT3*, *PTPN11*) and regulators of chromatin state (e.g., the histone 3 lysine 18 and 27 acetylase and transcriptional coregulator *CREBBP*, and the H3K36 methylase *WHSC1*) are frequent concomitant genetic events in high hyperdiploid ALL [23]. *CREBBP* mutations are enriched in the histone acetyl transferase domain and are selected during disease evolution [25]. As a potential mechanism for the generation of aneuploidy, hyperdiploid ALL blasts show a delay in early mitosis at prometaphase associated with defects in chromosome alignment, which lead to chromosome-segregation defects and nonmodal karyotypes [26]. Moreover, condensin complex activity is impaired, leading to chromosome hypocondensation, loss of centromere stiffness, and mislocalization of the chromosome passenger complex proteins Aurora B kinase (AURKB) and BIRC5 (survivin) in early mitosis [26]. Notwithstanding the favorable outcome of this subtype, condensin impairment suggests novel molecular targets (condensin-complex members, AURKB, or the spindle assembly checkpoint) for potential pharmacological intervention.

Hypodiploid ALL includes near haploid (24–31 chromosomes) and low hypodiploid (32–39 chromosomes) subtypes [27]. Near haploidy is present in ~2% of childhood ALL and is associated with Ras mutations (particularly *NF1*) and deletion/mutation of *IKZF3.* The gene expression profile and patterns of co-mutation (e.g., *CREBBP* and the Ras signaling pathway) are similar to high hyperdiploid ALL, suggesting a potential common origin of these two forms of leukemia. Low hypodiploidy instead is uncommon in children (~1%) but present in >10% of adults, and is characterized by deletion of *IKZF2*, *RB1*, *CDKN2A/CDKN2B* and near-universal mutations of *TP53* mutations, which are inherited in approximately half of cases and indicate that low hypodiploid ALL is a manifestation of Li-Fraumeni syndrome [28]. Duplication of the aneuploid genome, resulting in clones with 50 to 78 chromosomes, is common, with duplicated subclones present in the majority of cases. Predominance of the duplicated clone, known as masked hypodiploidy, may be misdiagnosed as high hyperdiploidy [29]. However, these states may usually be distinguished as the duplicated hypodiploid genome typically exhibits diploid and tetraploid chromosomes; in contrast high hyperdiploidy is characterized by a mixture of triploid and some tetraploid chromosomes (e.g., chromosomes 21, X). Moreover, the pattern of chromosomal losses in hypodiploid ALL is not random and chromosome 21 is never lost indicating a central role in leukemic cell fitness [27]. Hypodiploid ALL is associated with unfavorable outcome, although this is mitigated by minimal residual disease (MRD) risk-stratified therapy in several studies [30]. Moreover, for patients who achieve MRD-negative status after induction, allogeneic transplantation has been shown to be not successful in improving overall survival [31,32]. Although MRD-oriented protocols, older adults and elderly patients with low hypodiploidy do fairly poor with higher five-year cumulative incidence of relapse compared to high hypodiploid cases [33], making them candidates for different treatment approaches (e.g., immunotherapy and targeted therapies). Among those, preclinical studies have shown that hypodiploid ALL cells are sensitive to Phosphoinositide 3-kinase (PI3K) and BCL2 Apoptosis Regulator (BCL2) inhibitors [27,34].

#### 2.1.2. iAMP21

Intrachromosomal amplification of chromosome 21 (iAMP21) accounts for 1% of childhood ALL and is associated with older children (median age at diagnosis 9 years) and a low white cell count. Behind the formation of this chromosomal abnormality there is a characteristic mechanism of breakage–fusion–bridge cycles followed by chromothripsis and other complex structural rearrangements of chromosome 21 [35]. Two germline genomic alterations are associated with a markedly elevated risk of iAMP21. These are a germline Robertsonian translocation rob (15;21) and a germline ring chromosome 21 [36]. The presence of iAMP21 is associated with poor prognosis in most studies, although this has been improved with intensive treatment [37].

#### 2.1.3. Subtypes with Recurrent Chromosomal Translocations and/or Gene Fusions

The t(12;21)(p13;q22) translocation with the *ETV6-RUNX1* (*TEL-AML1*) fusion is the most common alteration in childhood B-ALL occurring in 20–25% of cases [38–40]. The *ETV6-RUNX1* fusion is considered to be a leukemia-initiating alteration which arises in utero, as demonstrated by the identification in umbilical cord blood [39] and by the prenatal monoclonal origin in identical twins [41]. The typically prolonged latency from birth to clinically manifest leukemia indicates that *ETV6-RUNX1* alone requires cooperating genetic events to induce leukemia, including deletion of the non-rearranged *ETV6* allele, focal deletion of *PAX5* and mutation of *WHSC1* [39,42–44].

The t(1;19)(q23;p13) translocation encoding *TCF3-PBX1* defines a subtype of 5–6% of pediatric B-ALL but only 1% of adult cases. This fusion is associated with a pre-B immunophenotype and expression of cytoplasmic immunoglobulin heavy chain and with higher peripheral blood white cell count at diagnosis [6,7,45]. Current intensive treatment has changed the historically high risk of *TCF3-PBX1* childhood ALL that was in part ascribed to central nervous system (CNS) involvement and relapse in favorable and intermediate risk cases [46,47]. *TCF3-PBX1* leukemic cells may be amenable to inhibition of pre-BCR signaling by dasatinib and ponatinib [48,49]. This approach may lead to compensatory upregulation of *ROR1* expression, and thus, concomitant inhibition of ROR1 could enhance the sensitivity of dasatinib [50]. *TCF3* and *TCF4* are also rearranged to *HLF*, and define a rare subtype of ALL (<1%) associated with an extremely poor prognosis [3,7]. *TCF3-PBX1* and *TCF3-HLF* ALL have distinct gene expression profiles and

mutational landscapes [7,51]. *TCF3-HLF* ALL is associated with expression of stem cell and myeloid markers, alterations of *PAX5* (deletions) and the Ras signaling pathway [7,51] and sensitivity to therapies inhibiting BCL2 and the pre-B cell receptor [52,53], immunologic therapies [54], and to Aurora A kinase inhibitors [55].

Rearrangements of the mixed-lineage leukemia 1 (*MLL1*) gene (now renamed Lysine [K]-specific methyl transferase 2A or *KMT2A)* on chromosome 11q23 to over 80 different partner genes define a subtype of leukemia with lymphoid and myeloid features and poor prognosis [12,56]. It occurs predominantly in infants (~80%), with a second peak of onset in adulthood where the most common partner of rearrangement is *AFF1* [57]. It is typically associated with pro-B (CD10-) immunophenotype, and expression of myeloid markers. Irrespective of fusion partner or lineage phenotype this subtype shows a distinct gene expression signature with overexpression of *HOX* cluster genes and the HOX cofactor *MEIS1* [58,59]. In infant *KMT2A*-rearranged ALL, the PI3K and Ras pathways are commonly altered [7,60,61]. *KMT2A* rearrangement is associated with altered chromatin patterning including H3K79 methylation, which has stimulated development of novel therapeutic approaches including inhibition of DOT1L [62], bromodomain, Menin, and the polycomb repressive complex [57,63,64]. The lineage plasticity characteristic of *KMT2A*rearranged ALL is important in the context of immunotherapy, as this may facilitate loss of expression of CD19 and escape from CD19 Chimeric antigen receptor T cell (CAR-T) therapy [65].

The frequency of patients with *BCR-ABL1* (Philadelphia chromosome) arising from the t(9;22)(q34;q11) translocation increases with age with 2–5% in childhood, 6% in adolescents and young adults (AYA), and more than 25% in adults [66,67]. Although historically considered a high-risk subtype, the incorporation of tyrosine kinase inhibitors (TKIs) into the standard treatment regimen for *BCR-ABL1*-positive ALL significantly improved clinical outcomes [68–70]. Secondary cooperative mutations are *IKZF1*, *PAX5* and *CDKN2A/B* deletions [42,69,71,72]. *IKZF1* alterations (most commonly deletions) have been associated with unfavorable outcome irrespective of TKI exposure [68,73], especially when co-occurring with (*CDKN2A* or *CDKN2B*, *PAX5*, or both: *IKZF1*plus) [68,69].

#### *2.2. Emerging B-ALL Subtypes Defined by Genome Sequencing Studies*

NGS approaches, particularly whole transcriptome sequencing (WTS), have enabled several research groups the identification of a large number of novel genetic alterations. These include cryptic rearrangements not identifiable by conventional approaches; novel subtypes that "phenocopy" established subtypes sharing similar gene expression profile but having different founding alterations; and subtypes defined by a single point mutation.

#### 2.2.1. *DUX4*, *MEF2D*, *ZNF384* and *NUTM1* Gene Fusions

Translocation of *DUX4* to the immunoglobulin heavy chain locus (*IGH*) is a cytogenetically cryptic alteration occurring in 5–10% of B-ALL and resulting in overexpression of a 3 truncated DUX4 protein [7,13,74–77]. *DUX4* is located within the D4Z4 subtelomeric repeat element on chromosome 4q/10q and encodes a double homeobox transcription factor that activates expression of large number of genes in early developing embryos, but it is thereafter silenced in most somatic cells [78]. Aberrant *DUX4* expression is associated with facioscapulohumeral dystrophy (FSHD) [79], while *DUX4* rearrangements have been also identified in Ewing-like sarcoma [80] and rhabdomyosarcoma [81]. In B-ALL, truncated DUX4 protein binds to an intragenic region of *ERG* resulting in transcriptional deregulation, and commonly, expression of a C-terminal ERG protein fragment, and/or *ERG* deletion. This subtype has a very distinctive gene expression profile and immunophenotype (CD2 and CD371 positive), common deletions of *IKZF1* (40%) and despite this, excellent outcome [76,77,82,83]. Accurate identification of all cases of *DUX4*-rearranged ALL requires direct identification of rearrangement (e.g., by WTS), or alternatively, gene expression-based clustering or high *DUX4* expression. The detection of strong CD371 cell surface expression by flow cytometry is a promising surrogate marker for this subtype [84]. Although *ERG* deletion is common in, and largely restricted to *DUX4*-rearranged ALL, these deletions are secondary events, commonly subclonal, and not present in all cases. Thus, the use of *ERG* deletion as a surrogate for identification of *DUX4*-rearranged ALL is suboptimal and should be avoided.

*MEF2D* (myocyte enhancer factor 2D) rearrangements occur in ~4% of childhood and 10% adult B-ALL cases. This subtype shows a distinct immunophenotype with low/absent expression of CD10, and positivity for CD38 and cytoplasmic μ chain, and distinct expression profiles [7,85–88]. *MEF2D* is the 5 partner in all described fusions, whereas B-cell CLL/lymphoma (*BCL*) 9 and heterogeneous nuclear ribonucleoprotein U-like 1 (*HNRNPUL1*) are the two most recurrent 3 partners. The rearrangements result in enhanced MEF2D transcriptional activity, increased *HDAC9* expression and sensitivity to histone deacetylase inhibitors, such as panobinostat [85]. MEF2D has also been implicated in a core transcription factor regulatory circuit involving SREBF1 that regulates pre-BCR and lipid metabolism, that are therapeutic vulnerabilities [89]. Sensitivity to staurosporine and venetoclax has been also described [90]. *MEF2D*-rearranged ALL shows high levels of minimal residual disease and is considered to be an unfavorable subtype because of its poor event-free survival rates [82,83].

*ZNF384-*, or less commonly, *ZNF362*-rearranged acute leukemia is a biologically and clinically distinct leukemic subtype present in ~6% of childhood, 7.3% of adult, and 15% of AYA B-ALL, and in 48% of B/myeloid mixed phenotype acute leukemia (MPAL) [7,13,91–93]. These cases show a characteristic immunophenotype with weak CD10 and aberrant expression of the myeloid markers, CD13, and/or CD33 [92,94]. Expression of myeloperoxidase (MPO) is often the only feature distinguishing cases diagnosed as B-ALL (MPO−) or B/myeloid MPAL (MPO+). Different fusion partners, usually transcription factor (e.g., *TAF15* and *TCF3*) or chromatin modifiers (e.g., *CREBBP*, *EP300*, *SMARCA2*, and *ARID1B*) have been identified for ZNF384, with EP300 being the most common. In all rearrangements the zinc-finger domains of the C2H2-type zinc-finger transcription factors ZNF384/ZNF362 are retained [7,13,91,92,95]. The same cooperating genetic alterations and transcriptional profile is observed in *ZNF384*-rearranged B-ALL and MPAL, and both exhibit lineage plasticity during disease progression (e.g., with shift in immunophenotype from lymphoid to myeloid from diagnosis to relapse). *ZNF384* rearrangements are acquired in a subset of hematopoietic stem cells and prime leukemia cells for lineage plasticity [92]. A report of *ZNF384*-rearranged ALL in twins implicated a fetal hematopoietic progenitor as the cell of origin confirming that these rearrangements are founder alterations [96]. Prognosis varies by fusion partner: the *EP300-ZNF384* fusion is associated with favorable outcome while the *TCF3-ZNF384* fusion is frequently associated with late relapses and a poor prognosis [92,93]. However, overexpression of *FLT3*, characteristic of this subtype, makes this leukemia amenable to FLT3 inhibition [97].

*NUTM1* (nuclear protein in testis midline carcinoma family 1) rearrangements (<2% of childhood B-ALL and mostly infant without *KMT2A*-rearrangements) [7,13,88,98–100] are characterized by fusion of *NUTM1* to different partners, including transcription factors and epigenetic regulators (e.g., *ACIN1*, *AFF1*, *ATAD5*, *BRD9*, *CHD4*, *CUX1*, *IKZF1*, *RUNX1*, *SLC12A6*, and *ZNF618*), that drive aberrant *NUTM1* expression [7,13]. In all fusions, the NUT domain is retained, and this is hypothesized to lead to global changes in chromatin acetylation [101] and to sensitivity to histone deacetylase inhibitors or bromodomain inhibitors in case of fusions with BRD9. *NUTM1* rearrangements confer an excellent prognosis to current therapeutic approaches [82,83,98]. Since not all *NUTM1* fusions are detectable by karyotyping either break-apart FISH or, preferably, WTS are the best approaches for diagnosis. In addition, the finding that both RNA expression of the 3 exons and protein expression are highly specific for this subtype may help in diagnosis.


**Figure 1.** This schematic algorithm for B-ALL subtyping was modified from the figure originally published in Paietta E. et al. Molecular Classification Improves Risk Assessment in Adult BCR-ABL1-negative B-ALL. Blood Prepublished Apr 25 2021; doi:10.1182/blood.2020010144 [83]. This figure describes each B-ALL subtype according to the specific genetic alterations and gene expression profile. Moreover, for each subtype peak prevalence and prognosis are shown. Subtypes are colored according to defining genetic alteration: gross chromosomal abnormalities (purple), transcription factor rearrangements (blue), other transcription factor alterations (blue), and kinase alterations (orange). Abbreviations: AYA, adolescent and young adult; tSNE, t-distributed stochastic neighbor embedding; TBD: to be defined; -R: rearranged.

#### 2.2.2. Subtypes That Phenocopy Established Subtypes Ph-Like ALL

Ph-like or *BCR-ABL1*-like ALL is characterized by a gene expression signature similar to that of Ph-positive ALL but lacking the pathognomonic BCR-ABL1 oncoprotein of Ph+ ALL [15,66,102–110]. Its incidence ranges from ~10–15% in children to ~20% in older adults, with a peak (25–30%) in the AYA ALL population. Similar to patients with Ph+ ALL, patients with Ph-like ALL often exhibit adverse clinical features and poor outcome and frequently harbor alterations of *IKZF1* or other B-lymphoid transcription factor genes. Over 60 heterogenous genetic alterations in kinases and cytokine receptors drive constitutively active kinase or cytokine receptor signaling, many of which have been shown to be druggable with a variety of kinase inhibitors. The most commonly mutated pathways are the ABL and JAK-STAT pathways with multiple rearrangements and lesions that converge on downstream ABL/JAK-STAT signaling. Founder alterations may be grouped into three broad types: (i) JAK/STAT alterations including: mutations activating cytokine receptors (e.g., *CRLF2* and *IL7R*); enhancer hijacking gene rearrangements deregulating cytokine receptor expression (e.g., *IGH-CRLF2* and *P2RY8–CRLF2*) [111–114]; gene fusions and/or mutations activating kinases (e.g., *JAK1*, *JAK2*, *JAK3*, *TYK2*); and rearrangements hijacking and truncating cytokine receptor expression (e.g., cryptic *EPOR* rearrangements) [115]; (ii) fusions involving ABL-class genes (*ABL1*, *ABL2*, *CSF1R*, *LYN*, *PDGFRA*, *PDGFRB*); and (iii) less common fusions (*FLT3*, *FGFR1*, *NTRK3*, *PTK2B*) [109] whose number is growing with increasing sequencing studies of different cohorts. Among these, alterations of *CRLF2* are present in approximately half of Ph-like ALL in AYAs and adults. *CRLF2* is located in the pseudoautosomal region of the sex chromosomes (*PAR1*) at Xp22.3/Yp11.3, and its alterations include: (1) a cryptic rearrangement that juxtaposes *CRLF2* to the IGH locus; (2) a focal deletion in the pseudoautosomal region of the sex chromosomes resulting in P2Y receptor family member 8 (*P2RY8*)-*CRLF2* fusion that positions *CRLF2* under the control of the *P2RY8* promoter; and less frequently by (3) an activating *CRLF2* point mutation, F232C. PAR1 deletions, as a surrogate for rearrangement of *CRLF2*, have been incorporated into the criteria for "*IKZF1*plus", a designation based on DNA copy number profiling, commonly by single nucleotide polymorphism (SNP) or multiplex ligation-dependent probe amplification (MLPA) arrays. In some studies, *IKZF1*plus has been associated with a higher risk of relapse defined by co-occurrence of the *IKZF1* deletion with deletion of *CDKN2A*, *CDKN2B*, *PAX5*, and/or PAR1 region in the absence of *ERG* deletion [116]. Notably, however, the *IKZF1*plus designation typically does not consider cases with *IGH-CRLF2* due to the inability of these platforms to detect this alteration.

The heterogeneous genomic landscape and often cytogenetically cryptic alterations identified in Ph-like ALL may make diagnosis of this entity and its driver alterations challenging, but several tractable diagnostic approaches are available, depending on technical capability of a laboratory, and the desired clinical/diagnostic endpoint (e.g., identification of the gene expression profile of Ph-like ALL c.f. identification of the most common driver kinase alterations). Comprehensive clinical NGS, including WTS, is the best approach to identify Ph-like ALL patients with targetable kinase alterations, as it enables analysis of gene expression, fusions, aneuploidy and sequence mutations. Selective/capture-based sequencing approaches (e.g., Archer FusionPlex, and FoundationOne Heme) also identify the majority of kinase-deregulating rearrangements in Ph-like ALL. If genomic approaches are not available, a more targeted screening approach using routine diagnostics, including flow cytometry (especially for *CRLF2*, for which positivity on flow cytometry is strongly correlated with rearrangement) and FISH for the most common kinase targets of rearrangement, is still effective for swift identification of Ph-like ALL [104]. The identification of specific genetic lesions is important for guiding targeted therapeutic intervention as a proportion of kinase-activating alterations in Ph-like ALL can, at least based on in vitro and preclinical models, be targeted by FDA-approved TKIs: JAK-STAT signaling (JAK inhibition); ABL-class fusions (ABL inhibitor); FLT3 and NTRK3 fusions (FLT3 and NTRK3 inhibitor) [104]. Several recent studies have described the efficacy of

ABL1 and NTRK inhibitors in the treatment of Ph-like ALL cases with rearrangement of these genes [117,118]. Combinatorial use of kinase inhibitors against multiple signaling has shown synergism in patient-derived xenograft (PDX) models of CRLF2/JAK mutant (JAK and PI3K/mTOR inhibitors), ABL/PDGFR mutant (dasatinib and PI3K/mTOR inhibitor) and EPOR-rearranged (ponatinib and ruxolitinib) [119]. Moreover, recently dual JAK/GSPT1-degrading proteolysis-targeting chimeras PROTACs have been developed and showed efficacy in Ph-like B-ALL kinase-driven PDX models which were otherwise unresponsive to type I JAK inhibitors [120]. Lastly, the use of immunotherapeutic agents, such as blinatumomab, inotuzumab, and CAR-T cells (including those targeting CRLF2 [121], represents a promising alternative approach for this subtype which is irrespective of a specific genetic alteration or response to prior chemotherapies [104]).

#### *ETV6*-*RUNX1*-like ALL

*ETV6*-*RUNX1*-like ALL is characterized by a gene expression profile and immunophenotype (CD27 positive, CD44 low to negative) similar to *ETV6*-*RUNX1* ALL, but lacking the *ETV6*-*RUNX1* fusion [122] and favorable prognosis [7,13,75,122,123]. These cases harbor alternate gene fusions or copy number alterations in ETS family transcription factors (*ETV6*, *ERG*, *FLI1*), *IKZF1* or *TCF3. ETV6-RUNX1*-like ALL develops in children harboring germline *ETV6* mutations with subsequent somatic alterations of the second *ETV6* allele, consistent with the notion that biallelic alteration of *ETV6* is central in leukemogenesis [124]. It is more common children (~3%) and confers an unfavorable prognosis in children due to high levels of MRD and worst event-free survival rates [82].

#### 2.2.3. Subtypes Defined by a Single Point Mutation

#### PAX5 P80R and PAX5alt

The PAX5 P80R subtype (~3% of B-ALL cases) is characterized by the presence of a hot spot mutation at amino acid 80 in the DNA binding domain of the paired box DNA-binding transcription factor PAX5 [7,13,125,126]. B-ALL cases with PAX5 P80R show a distinct gene expression profile with the majority of cases having either hemizygous or homozygous mutation, caused by deletion of the wild-type *PAX5* allele or copy-neutral loss of heterozygosity. In a subset of cases, in addition to PAX5 P80R there is a second frameshift, nonsense or deleterious missense *PAX5* mutation. Thus, biallelic *PAX5* alterations—with mutation of one allele and loss of activity of the second allele—are a hallmark of this subtype [7,125]. In support of the role of biallelic alteration of *PAX5* in the pathogenesis of this subtype, knockin mouse models of germline *Pax5* mutations have shown that heterozygous Pax5P80R/+ knock-in mice develop transplantable B-ALL, with genetic inactivation of the wildtype *Pax5* allele [7]. In contrast, Pax5G183S/+ knock in mice (modeling the germline PAX5 G183S mutation observed in familial ALL) show a low penetrance of ALL [127], supporting its role as a haploinsufficient tumor suppressor. Additional important cooperating lesions include *CDKN2A* loss and signaling pathway mutations, most commonly in Ras signaling genes or in the JAK/STAT pathway [7,125].

PAX5alt comprises about 7% cases with diverse *PAX5* alterations, including rearrangements, sequence mutations and focal intragenic amplifications [7]. Over 20 different partner genes have been identified with *PAX5-ETV6* being the most common. Children in this subtype are more commonly classified as high risk rather than standard risk (according to National Cancer Institute (NCI) criteria). In addition to *PAX5* alterations, recurrent genetic lesions observed in these cases include copy number losses affecting cell-cycle regulation genes such as *CDKN2A*, *RB1* and *BTG1*, B-cell development genes, transcriptional regulators and/or epigenetic modifiers (for example, *KDM6A*, *KMT2A* and *ATRX*) [7]. Both PAX5 P80R and PAX5alt subtypes are associated with intermediate to favorable prognosis [7,82,83,126].

#### IKZF1 N159Y

*IKZF1* encodes the transcription factor IKAROS, which is a member of the family of zinc finger DNA-binding proteins required for lymphoid lineage ontogeny and homeostasis [128,129]. The most common type of *IKZF1* alteration is a focal deletion occurring in 15% of ALL cases and in >50% of high risk ALL [42,72,103]. Deletions result in loss-of-function or in the dominant negative IK6 isoform and are associated with poor outcome [73,103]. In addition to deletions, missense, frameshift and nonsense mutations have been also described in pediatric high-risk B-ALL patients. Among those the missense p.Asn159Tyr mutation defines a subtype (<1% of B-ALL) with a distinct gene expression profile characterized by upregulation of genes with roles in oncogenesis (the IKZF1-interacting gene *YAP1*), chromatin remodeling (*SALL1*), and signaling (*ARHGEF28*) that are not deregulated in other subgroups of *IKZF1*-altered ALL [7,74]. In contrast to PAX5 P80R ALL, the nonmutated wild-type allele of the mutated transcription factor (here *IKZF1*) is retained [72]. As for most other missense mutations observed in IKZF1 zinc fingers, IKZF1 N159Y induces misregulation of IKZF1 transcriptional activation, in part through distinctive nuclear mislocalization and enhanced intercellular adhesion [130].

#### ZEB2 H1038R and IGH-CEBPE

In unsupervised clustering of leukemic cell gene expression, cases with the H1038R mutation in *ZEB2* phenocopy the translocation t(14;14)(q11;q32) [13], which results in IGH-CEBPE fusion, suggesting a common activated pathway of leukemogenesis and defining a rare B-ALL subtype (<1%). This is associated with *NRAS* sequence mutations (>50% of cases), upregulation of *LMO1* and downregulation of *SMAD1* and *BMP2* [10]. However, neither the *IGH* or *ZEB2* mutations are unique to this group, nor do they explain all cases in this distinct gene expression and experimental validation is required to demonstrate their role as leukemogenic drivers. B-ALL with *ZEB2* mutation is associated with poor event-free survival and high relapse [131].

#### *2.3. Prognostic Implications*

The frequent and wide use of genomics to profile the landscape of ALL has allowed a tailored refinement of risk in association with standard criteria, such as MRD levels [82] (Figure 1). In childhood B-ALL, *ETV6-RUNX1*, high-hyperdiploid, and *DUX4*-rearranged B-ALL are categorized as favorable due the highest overall survival rates and the lowest relapse rates, despite elevated early MRD in *DUX4*-rearranged cases. *BCR-ABL1*, *BCR-ABL1*-like, *ETV6-RUNX1*-like, *KMT2A*-rearranged, and *MEF2D*-rearranged ALL show high levels of MRD and the worst event-free survival rates and thus are categorized to be unfavorable subtypes. The remaining subtypes including *TCF3-PBX1*, *PAX5*alt, iAMP21, hypodiploid, *ZNF384*-rearranged, *NUTM1*-rearranged, and *PAX5* P80R ALL have intermediate risk [81]. These prognostic groups have been mostly confirmed in a historic, non-MRD risk adapted trial (UKALLXII/ECOG-ACRIN E2993, NCT00002514) in adolescents and adult B-ALL cases [83] according to the following risk assignment: standard risk genotypes: *DUX4*-rearranged, *ETV6-RUNX1*/-like, *TCF3-PBX1*, PAX5 P80R, highhyperdiploid; high-risk genotypes: Ph-like, *KMT2A-AFF1*, low-hypodiploid/near-haploid, BCL2/MYC-rearranged; and intermediate-risk genotypes: *PAX5*alt, *ZNF384*/-like, and *MEF2D*-rearranged.

#### **3. T-Cell Acute Lymphoblastic Leukemia (T-ALL)**

#### *3.1. Genomic Overview of T-ALL*

T-ALL leukemic cells express a subset of T-cell makers (CD3, cyCD3, CD2, CD5, CD7, CD8) and arises from immature T-cell progenitors [132,133]. Pediatric T-ALL accounts for 10–15% of newly diagnosed pediatric ALL and is characterized by higher incidence in boys, high initial white blood cell counts, mediastinal mass, CNS infiltration, and slightly worse prognosis compared to B-ALL [134]. The majority of T-ALL cases may be subclassified into subtypes according to the aberrant expression and dysregulated

pathways of transcription factors and oncogenes induced by leukemia-initiating alterations involving basic helix–loop–helix (bHLH) factors (*TAL1*, *TAL2*, *LYL1*), homeobox genes (*TLX1* (*HOX11*), *TLX3* (*HOX11L2*), *NKX2-1*, *NKX2-5*, *HOXA*), *LMO1*, *LMO2*, *MYB*, *BCL11B* and *SPI1* (Figure 2) [5,135,136]. These subtypes are defined with expression profiles by WTS or microarray, however, almost half of these leukemia-initiating alterations in T-ALL show intergenic breakpoints that can be missed by WTS but rescued by whole genome sequencing (WGS) [135,136]. Epigenomic analyses have also identified novel leukemiainitiating alterations in non-coding regions [5,137,138].

**Figure 2.** T-cell differentiation and T-ALL subtyping. This schema describes differentiation stages of each T-ALL subtype according to the specific genetic alterations leading to aberrant expression of rearranged or mutated genes. Prevalence and prognosis of each subtype are shown. Subtypes are colored according to corresponding normal T-cell differentiation stage: early T-cell precursor (ETP, red and orange), early stages of cortical thymocytes maturation (green), and late stages of cortical thymocytes maturation (blue). Abbreviations: T-ALL: T-cell acute lymphoblastic leukemia; T/M MPAL: T/myeloid mixed phenotype acute leukemia; -R: rearranged.

*NOTCH1* activating mutations and deletion of *CDKN2A/CDKN2B* loci (9p21) are found in over 70% of T-ALL cases and considered as secondary but core events in leukemogenesis [135,136,139]. Concurrent somatic mutations and copy number alterations are frequently observed in T-ALL leading to dysregulation of several cellular pathways, including JAK-STAT signaling (*IL7R*, *JAK1*, *JAK3*, *DNM2*), Ras signaling (*NRAS*, *KRAS*, and *NF1*), PI3K-AKT signaling (*PTEN*, *AKT1*, *PIK3CA PIK3CD*), epigenetic regulation (*PHF6*, *SUZ12*, *EZH2*, *KDM6A*), transcription factors and regulators (*ETV6*, *GATA3*, *RUNX1*, *LEF1*, *WT1*, *BCL11B*), and translation regulators (*CNOT3*, *RPL5*, *RPL10*) [135,136,140,141]. Accumulation of these aberrant expression and dysregulated pathways disrupt the normal T-cell differentiation, proliferation, and survival, and results in T-ALL with unique gene expression signatures reflecting the point of differentiation arrest during T-cell development [133,140]. In addition to expression profiles, DNA methylation signatures are also associated with immunophenotypic profiles and normal T-cell development differentiation stage [142,143].

#### *3.2. T-ALL in Early Stages of Cortical Thymocyte Maturation*

T-ALL with CD1a+, CD4+, and CD8+ immunophenotype includes several subgroups, such as rearrangements of *TLX1*, *TLX3*, *NKX2-1*, reflecting a differentiation arrest in early stages of cortical thymocyte maturation and confers a relatively favorable prognosis [144,145]. These subgroups almost commonly harbor *NOTCH1* and *CDKN2A* alterations. Dysregulated expression of HOX transcription factor genes is mostly induced by chromosomal translocations and inversions that juxtapose these genes to enhancers in the *TCR* and

*BCL11B* regulatory regions [135,146]. Importantly, *BCL11B* rearrangements (*BCL11B*-*TLX3*) in this subgroup are mechanistically distinct from those identified in *BCL11B*-rearranged lineage ambiguous leukemias, in that in the *BCL11B-TLX3* leukemia, the *BCL11B* enhancer is used for aberrant expression of *TLX3* at the cost of the loss of expression of *BCL11B*, leading to complete difference in expression profiles [5,146,147]. Instead, *TLX3* rearranged T-ALL (including *BCL11B*-*TLX3*) shares gene expression signatures, DNA methylation profiles, somatic mutations (*BCL11B*, *WT1*, *PHF6*, *DNM2*), and downstream targets (JAK-STAT, epigenetic regulators) with *TLX1* rearranged T-ALL [135,136,142,143,145]. Some of overlapping genomic features with *TLX1*/*TLX3* rearranged T-ALL, including *NUP214-ABL1* (TKIs) and JAK-STAT pathway (ruxolitinib, a JAK-STAT inhibitor), can be targetable and have been incorporated into ongoing clinical trials [148].

#### *3.3. TAL1-Driven T-ALL with Late Stages of Cortical Thymocyte Maturation*

Deregulation of the *TAL1* oncogene is a feature of T-ALL that typically exhibits a late cortical thymocyte immunophenotype (CD4+, CD8+, CD3+) and comprises approximately 40% of T-ALL [135,136]. This T-ALL subtype includes *TAL1* and *TAL2* rearranged cases and is further classified into two subgroups by expression profiles whose one expresses *PTCRA* (pre-TCR) suggesting LCK activation that correlated with dasatinib sensitivity [136,145,149]. During normal T-cell differentiation, *TAL1* expression is transcriptionally silenced along with T-cell lineage commitment to proceed appropriate *TCR* rearrangements and differentiation [133]. *TAL1* overexpression is induced by several mechanisms: (1) chromosomal translocations with *TCRA/D*; (2) sub-microscopic interstitial deletion (*STIL-TAL1*); (3) disruption of insulated neighborhoods by losing CTCF binding sites [150]; and (4) somatic indels in a noncoding intergenic regulatory element upstream of *TAL1* to generate aberrant MYB binding site (*MuTE*) [137]. The latter two mechanisms have benefited of NGS technologies for their identification. Dysregulated *TAL1* expression inhibits the function of E-protein dimers by forming TAL1-E-protein heterodimer [151]. Furthermore, TAL1 forms the central node of the core regulatory circuit to coordinately regulate downstream target genes with several hematopoietic transcription factors including *GATA3*, *RUNX1*, *MYB*, and the ETS family genes, which is active in normal hematopoietic stem cell (HSC) and progenitor cells [152,153], and RUNX1 inhibition is reported to impair the growth of T-ALL but not normal hematopoietic cells [154]. However, although TAL1 functions as a master transcription factor related to T-cell differentiation and leukemogenesis of T-ALL, only 30% of transgenic mice develop T-ALL after a latent period, indicating that additional abnormalities are required for leukemogenesis [155]. Expression of *Lmo2* accelerates the onset of leukemia in *Tal1* transgenic mice, and *LMO1/LMO2* are commonly expressed in human *TAL1*-driven T-ALL [156,157]. Other cooperative genes and noncoding RNAs in *TAL1*-driven T-ALL include *ARID5B*, *ARIEL*, and *MYC*, driving aberrant expression of *TAL1* [158,159]. In addition, PI3K-AKT pathway genes including *PTEN* are frequently mutated in this subgroup [135,136], which associates with glucocorticoid resistance and can be reversed by the inhibition of this pathway [160]. Several cell cycle regulators including *CDK6* and *CCND3* are regulated by TAL1 complex [152] and may be potential targets of therapeutic intervention [161].

#### *3.4. Early T-Cell Precursor (ETP) ALL and Mixed Phenotype Acute Leukemia*

ETP-ALL is often referred to as a subtype of T-ALL as it exhibits an immunophenotype analogous to the earliest stages of T-cell development (cytoplasmic CD3+, CD7+; CD8−, CD1a−, CD5weak), and with expression of myeloid and/or stem-cell markers [144,162]. However, the genomic alterations and gene expression profile of ETP-ALL are more similar to a hematopoietic stem cell than a T cell precursor, suggesting that ETP-ALL could be included in a subgroup of immature acute leukemias of ambiguous lineage (ALAL), originating from a hematopoietic progenitor at a maturational stage prior to initiation of a definitive program of T cell differentiation. Consistent with this, recent studies have defined a subgroup of *BCL11B*-deregulated ALAL, that includes one third of ETP-ALL

and T/myeloid mixed phenotype acute leukemia (T/M MPAL) cases with a very distinct expression profile [5]. *BCL11B*-deregulated ALAL is characterized by structural variations of the region containing *BCL11B* at 14q32 including translocations and high-copy amplification generating a distal neo-enhancer, that each leads to aberrant expression of *BCL11B*, in the case of the rearrangements by hijacking super-enhancers active in CD34+ hematopoietic stem and progenitor cell (HSPCs) [5,147]. *FLT3* activating mutations were found in 80% of *BCL11B*-deregulated ALAL, and concurrent expression of *BCL11B* and FLT3-ITD on HSPC exhibited synergistic effects on activating T-cell directed differentiation to express cytoplasmic CD3 while blocking myeloid differentiation [5]. Other genomic features of ETP-ALL include a subgroup of aberrant expression of PU.1 (*SPI1* fusions), *HOXA* genes (rearrangements of *HOXA* genes, *KMT2A* rearrangements, *PICALM*-*MLLT10*, *SET*-*NUP214*) and mutations of multiple cellular pathways (Ras signaling, JAK-STAT signaling, and epigenetic regulators) and transcription factors related to T-cell development [135,136,163]. Especially, T-ALL with *SPI1* fusions represents unique expression profiles with high relapse rate [5,136]. Again, several of these genomic mutations were shared with T/M MPAL, including biallelic *WT1* alterations, mutations of hematopoietic transcription factors (*ETV6*, *RUNX1*, *CEBPA*) and activating mutations of signaling pathways (JAK-STAT, *FLT3*, Ras) [92,163], supporting that they are similar entities in the spectrum of immature leukemias and both might have sensitivity to FLT3 and/or JAK inhibition [164].

#### *3.5. NOTCH1 Activating Mutations in T-ALL*

*NOTCH1* encodes a highly conserved ligand-dependent transcription factor. The NOTCH1 signaling pathway plays an important role in the commitment of T-cell lineage specification and for further T-cell development [133,165]. In T-ALL, NOTCH1 activating mutations are found in more than 70% of cases and is considered an oncogene involved in leukemogenesis [135,136]. Aberrant activation of NOTCH1 pathway in T-ALL is mostly induced by (1) ligand-independent activation (somatic mutations, indels and large deletions that disrupt the negative regulatory region), or (2) impairment of the proteasomal degradation of intracellular domain of NOTCH1 (truncation of the PEST domain, *NOTCH1* mutations in 3 untranslated region, and *FBXW7* mutations) [166–171]. These two types of *NOTCH1* activating mutations have synergistic effects and more than 20% of T-ALL cases harbor both types of alterations [166]. However, most *NOTCH1* activating mutations found in human T-ALL are considered as a weak tumor initiator event. Co-existence of both types of *NOTCH1* mutations in hematopoietic progenitors tends to induce a transient preleukemic CD4+/CD8+ double positive cells and takes 10 to 15 weeks to fully transform into T-ALL, suggesting that they are alone incompletely leukemogenic [172–174]. In addition, more than 40% of T-ALL cases harbor subclonal *NOTCH1* activating mutations and their heterogeneity at diagnosis was reported by several studies [135,136,175]. Furthermore, *NOTCH1* activating mutations are considered to be acquired as a late secondary event in leukemogenesis [139,175,176].

A key target of NOTCH1 is the *MYC* oncogene that shares several overlapping target genes with NOTCH1 to promote cell proliferation and dysregulate anabolic pathways in T-ALL [174,177,178]. NOTCH1 controls T-cell-specific distal enhancer of *MYC* ("NMe"), resulting in the NOTCH1-MYC regulatory circuit [174,177,178]. In addition, pre-TCR signaling also correlates with NOTCH signaling, leading to LCK signaling and robust cell growth at DN3 stage in the T-cell development, which can be targetable by dasatinib [149,179].

Due to the high prevalence and importance of *NOTCH1* activating mutations in T-ALL, targeted therapy on NOTCH1 pathway has been a major interest. This includes γ-secretase inhibitors (GSIs), ADAM inhibitors, SERCA inhibitors, and monoclonal antibodies [180]. Among them, GSIs, that block the activation process of NOTCH receptors by inhibiting proteolytic cleavage, have been tested in preclinical and Phase 1 studies [181,182]. However, the usage of GSIs in T-ALL is still in a developing phase due to gastrointestinal toxicity and insufficient antitumor responses that mostly induce transient growth arrest rather

than cell death [183,184]. To overcome these problems, combination with other agents have been explored including glucocorticoids that showed synergistic effects by reversing glucocorticoid resistance [185]. Inhibition of mTORC1 signaling and PKCδ signaling are also promising combination strategies to restore GSIs sensitivity in resistant cells [186,187].

#### **4. Implications for Diagnosis**

The revolution in genomic characterization of ALL has created important opportunities and challenges for the clinical implementation of sequencing-based approaches for diagnosis and management of ALL (Table 1). This is particularly true for B-ALL, where many of the recently identified subtypes are associated with prognosis (even in the context of MRD-based risk-adapted therapy) [82,83] and where molecular characterization is needed to identify patients suitable for targeted therapy (an exemplar being Ph-like ALL). This is currently less compelling for T-ALL where identification of founding lesions driving T-ALL subtypes are of biological and mechanistic interest but are not typically used to risk stratify or guide therapy, exceptions possibly being kinase inhibition for JAK-STAT alterations and *ABL1* rearrangements, identification of alterations in Ras, *PTEN*, *NOTCH1* and/or *FBXW7* that have been found to be associated with outcome in some studies [188], and LCK dependence for dasatinib therapy [149]. The challenge is clinical implementation of appropriately comprehensive diagnostic approaches to identify all key genomic features. Despite the mutationally sparse genome of ALL, there is striking diversity of the nature of underlying driver alterations, including sequence mutations, DNA copy number alterations, and structural variations, many of which may involve the non-coding genome. Accurate subtyping is also challenged by the inability of conventional cytogenetic and targeted molecular approaches to identify several types of driver (e.g., *DUX4*-rearrangement) and the importance of identifying phenocopies (e.g., *ETV6-RUNX1*-like, and Ph-like ALL). Thus, moving forward, optimal clinical diagnostics require genomic approaches. The choice of approach in part rests on how clinical information will be used. If comprehensive subtyping and identification of all potentially clinically relevant genomic alterations is desirable, a combination of DNA and RNA-based technologies is required. For example, the combination of WGS and WTS enables the identification of sequence mutations, DNA copy number alterations, aneuploidy and structural variants (from WGS) together with identification of fusion chimeras, mutant allele expression, and gene expression profiling (from WTS). The use of one or both approaches is becoming increasingly widely used, and at St Jude Children's Research Hospital, three platform sequencing (WGS, WTS and exome sequencing) is clinical standard of care, informs clinical decision making in ALL [148], and retrieves more actionable clinical information than any single platform alone [189]. WGS is offered using a paired non-tumor sample to aid identification of somatic variants and provides the opportunity to return clinically relevant germline findings. Moreover, this comprehensive approach enables a more streamlined workflow [190–192], provided the demands of analysis and interpretation can be met.

However, WGS is not yet widely used clinically, and many clinicians and providers seek alternative approaches to identify clinically relevant alterations. These fall into three main categories: single platform sequencing, sub-genomic sequencing, and targeted detection of genomic alterations. In the first category, single platform WTS provides near comprehensive characterization of clinically relevant alterations in ALL, particularly B-ALL: gene expression-based profiling to identify subgroups and phenocopies; fusion transcripts; and interrogation of specific sequence mutations (e.g., JAKs, PAX5 and IKZF1) [7,193]. Moreover, several methods are available that utilize expression and mutant allele fraction to robustly identify large scale chromosomal copy number changes, thus providing a surrogate for conventional cytogenetic identification of aneuploidy [7,194]. WTS as a single platform has limitations—it is challenging to identify all sequence variations although analytic platforms are improving, it cannot identify focal DNA copy number alterations that may impact targetable pathways (e.g., *SH2B3* deletions in JAK-STAT-driven Ph-like ALL) and does not identify rearrangements that may deregulate oncogenes without resulting

in a RNA chimera—for example rearrangements of oncogenes in T-ALL such as *TLX3* and those involving TCR, where breakpoints are frequently intergenic [135], the diverse rearrangements in *BCL11B*-rearranged ALAL [5], or non-coding sequence mutations that drive oncogenes such as *TAL1* and *LMO1/2* [137,138].

Several platforms are available for targeted DNA and/or RNA sequencing, often using capture-based approaches. These including Foundation Medicine [195] and the FusionPlex ALL Kit (Invitae, previously ArcherDx). These have the advantage of being somewhat simpler to access or implement in routine diagnostic laboratories, and the ability to detect the majority of chimeric fusion events in B-ALL. Similar reservations to WTS apply regarding the limited ability of these platforms to detect intergenic rearrangements in ALL; moreover, these platforms either have limited (Foundation) or no (Archer) capability to detect DNA copy number alterations, particularly those that are single copy, and may have difficulty resolving complex rearrangements (e.g., truncating rearrangements of *EPOR* in Ph-like ALL) [115]. Capture based DNA sequencing for sequence mutations is widely used in hematological malignancies, but is not well suited to diagnosis of ALL due to the lack of detection of rearrangements and structural variations. As described above, the MLPA platform is widely used by several groups to identify focal DNA copy number alterations and the "*IKZF1*plus" composite genotype, but this platform is not an adequate surrogate for sensitive detection of several key subtypes: e.g., *ERG* deletion in *DUX4*-rearranged ALL (only ~50% of cases have clonal *ERG* deletion), and PAR1 deletion in *CRLF2*-rearranged ALL (*IGH-CRLF2* is usually not accompanied by PAR1 deletion).

**Table 1.** Clinical implementation of high-throughput sequencing.


The parenthesis in "Capability" indicates analyses in development. Abbreviations: WTS: whole transcriptome sequencing; BCR: B-cell receptor; TCR: T-cell receptor; WGS: whole genome sequencing; GWAS: genome wide association study; WES: whole exome sequencing; -R: rearranged.

In the absence of sequencing-based approaches, several subtypes and drivers may be identified by flow cytometry, immunophenotypic and targeted molecular approaches. Flow cytometry may be used to detect CRLF2 rearrangements, that result in cell surface expression of *CRLF2*, as well as markers associated with distinct subtypes (e.g., CD371 in *DUX4*-rearranged ALL). FISH may be used to detect rearrangement of the most commonly rearranged genes in Ph-like ALL for which targeted therapies are currently available (e.g., *ABL*-family kinase genes, *CRLF2*, *NTRK3*) with caveats—for example, the focal insertions of *EPOR* into *IGH* and similar enhancer regions are not robustly detected by FISH due to the small size of the *EPOR* insertion. Specific subtype-defining rearrangements may be detected by conventional molecular approaches such as RT-PCR. Thus, these composite approaches may be suitable to detect many actionable alterations in ALL, but do not provide a pathway to comprehensive identification of all driver lesions of prognostic significance.

#### **5. Conclusions**

Large-scale integrative genome-wide sequencing studies have profoundly transformed the molecular taxonomy of ALL, resulting in the identification of new entities with prognostic and therapeutic significance. There are over 30 different B/T-ALL subtypes defined by distinct constellations of somatic and/or germline genetic alterations that converge on distinct gene expression patterns. The identification of these dysregulated pathways is crucial for clinical management of ALL patients and most importantly for guiding therapeutic intervention. The best example is provided by the constitutively active kinases in Ph-like which are druggable by a variety of single or combinatorial TKIs. Although the enormous clinical and genetic progress of the past decade, much work remains, as most studies have lacked NGS and have not validated the mechanisms by which fusions/mutations cooperate in leukemogenesis, and not fully defined potential for targeting. Due to the heterogeneity of genetic lesions, optimal clinical diagnosis of ALL requires genomic and/or transcriptomic sequencing in order to identify fusions, aneuploidy and sequence mutations required for disease stratification. The use of such approaches is becoming increasingly widespread. Recently, new immunotherapeutic agents (e.g., developed antibodies and CAR-T cells) have been efficacious in a proportion of patients, but failed in others. Thus, efforts should be focused in the future on defining subtype specific vulnerabilities to improve treatment strategy and outcome.

**Author Contributions:** Writing—original draft preparation, I.I., S.K. and C.G.M.; writing—review and editing, I.I., S.K. and C.G.M.; visualization, I.I. and S.K.; supervision, I.I. and C.G.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** Work conducted by the authors described in this review was supported by the National Institutes of Health, including an NCI Outstanding Investigator Award R35 CA197697 (to C.G.M.), NCI Cancer Center Support Grant (CA021765) an St. Baldrick's Foundation Robert J. Arceci Innovation Award (to C.G.M.) and the Henry Schueler 41&9 Foundation (to C.G.M.); a Garwood Postdoctoral Fellowship of the Hematological Malignancies Program of the St Jude Children's Research Hospital Comprehensive Cancer Center (to S.K.); and the American Lebanese Syrian Associated Charities of St. Jude Children's Research Hospital.

**Acknowledgments:** The authors thank collaborators that have contributed patient samples, data, intellectual and technical expertise to the work described in this study, particularly the St. Jude Children's Research Hospital—Washington University Pediatric Cancer Genome Project, Children's Oncology Group Therapeutically Applicable Research to Generate Effective Treatments (TARGET) project (https://ocg.cancer.gov/programs/target/projects/acute-lymphoblastic-leukemia, accessed on 19 August 2021), the Eastern Cooperative Oncology Group—American College of Radiology Imaging Network (ECOG-ACRIN), The Alliance—Cancer and Leukemia Group B, and MD Anderson Cancer Center.

**Conflicts of Interest:** I.I. has received honoraria from Amgen and Mission Bio; S.K. declares no conflict of interest.; C.G.M. has received research funding from Loxo Oncology (relevant to the use of TRK inhibition in Ph-like ALL), AbbVie (venetoclax in ALL) and Pfizer; honoraria from Amgen and Illumina. These funders had no role in the writing of this manuscript.

#### **References**

