Case Report: Rare IKZF1 Gene Fusions Identified in Neonate with Congenital KMT2A-Rearranged Acute Lymphoblastic Leukemia

Chromosomal rearrangements involving the KMT2A gene occur frequently in acute lymphoblastic leukaemia (ALL). KMT2A-rearranged ALL (KMT2Ar ALL) has poor long-term survival rates and is the most common ALL subtype in infants less than 1 year of age. KMT2Ar ALL frequently occurs with additional chromosomal abnormalities including disruption of the IKZF1 gene, usually by exon deletion. Typically, KMT2Ar ALL in infants is accompanied by a limited number of cooperative le-sions. Here we report a case of aggressive infant KMT2Ar ALL harbouring additional rare IKZF1 gene fusions. Comprehensive genomic and transcriptomic analyses were performed on sequential samples. This report highlights the genomic complexity of this particular disease and describes the novel gene fusions IKZF1::TUT1 and KDM2A::IKZF1.


Introduction
The histone-lysine [K] Methyl Transferase 2A (KMT2A) on chromosome 11q23 is a pathogenic driver gene in acute lymphoblastic leukemia (ALL). KMT2A-rearranged ALL (KMT2Ar ALL) has poor long-term survival rates and is the most common ALL subtype in infants (<1 year of age) [1], comprising >70% of all ALL diagnoses in this age group [2]. It is often associated with hyperleukocytosis and central nervous system (CNS) involvement [3]. While KMT2Ar ALL frequently occurs with additional chromosomal abnormalities including disruption of the IKZF1 gene (chromosome 7p12), usually by exon deletion [4], KMT2Ar ALL in infants typically presents with few co-occurring alterations [5]. However, we report a genomically complex case of aggressive congenital KMT2Ar ALL harboring additional rare and novel IKZF1 gene fusions.
The infant was treated with induction chemotherapy as per the Interfant-06 protocol [6] and achieved a minimal residual disease (MRD) level of 5 × 10 −3 by the end of induction using Allele Specific Oligonucleotide PCR for IgH rearrangement. However, further intensive chemotherapy was delayed by significant toxicities, including vincristinerelated polyneuropathy requiring mechanical ventilation. Following two weeks of dosereduced mercaptopurine and oral methotrexate maintenance, azacitidine was administered (2.5 mg/kg for 5 days) [7] with the intention of bridging to hematopoietic stem cell transplant if remission was obtained. Unfortunately, the bone marrow blast percentage rose to 5% at the end of this cycle, and despite further azacitidine and consolidation chemotherapy as per the protocol AALL15P1 [7], by the next bone marrow examination, lymphoblasts had risen to 55%. Subsequent blinatumomab combined with intrathecal chemotherapy resulted in a morphological remission with MRD of 10 −2 . However, the second blinatumomab course was complicated by the development of facial nerve palsy secondary to an extradural leukemic chloroma impinging on her facial nerve. This heralded florid morphological relapse with 42% bone marrow blasts, which exhibited complete loss of CD19 on flow cytometry (93% CD19/34 + at diagnosis). The patient failed to respond to either FLAG-IDA or inotuzumab and succumbed at nine months of age. The timeline of treatments and response assessments are summarized in Figure 1, and immunophenotyping flow cytometric dot plots are shown in Figure S1. Cytogenetics demonstrated a 47XX karyotype with trisomy 8 and translocations t(4;11)(q21;q23) and t(7;11)(q11.2;p11.2). The KMT2A::AFF1 (MLL::MLLT2/AF4) KMT2A rearrangement was confirmed by fluorescence in situ hybridization. No diagnostic bone marrow biopsy was performed. A lumbar puncture revealed CNS involvement. The infant was treated with induction chemotherapy as per the Interfant-06 protocol [6] and achieved a minimal residual disease (MRD) level of 5 × 10 −3 by the end of induction using Allele Specific Oligonucleotide PCR for IgH rearrangement. However, further intensive chemotherapy was delayed by significant toxicities, including vincristine-related polyneuropathy requiring mechanical ventilation. Following two weeks of dose-reduced mercaptopurine and oral methotrexate maintenance, azacitidine was administered (2.5 mg/kg for 5 days) [7] with the intention of bridging to hematopoietic stem cell transplant if remission was obtained. Unfortunately, the bone marrow blast percentage rose to 5% at the end of this cycle, and despite further azacitidine and consolidation chemotherapy as per the protocol AALL15P1 [7], by the next bone marrow examination, lymphoblasts had risen to 55%. Subsequent blinatumomab combined with intrathecal chemotherapy resulted in a morphological remission with MRD of 10 −2 . However, the second blinatumomab course was complicated by the development of facial nerve palsy secondary to an extradural leukemic chloroma impinging on her facial nerve. This heralded florid morphological relapse with 42% bone marrow blasts, which exhibited complete loss of CD19 on flow cytometry (93% CD19/34 + at diagnosis). The patient failed to respond to either FLAG-IDA or inotuzumab and succumbed at nine months of age. The timeline of treatments and response assessments are summarized in Figure 1, and immunophenotyping flow cytometric dot plots are shown in Figure S1.   by mRNA-Seq are indicated. Asterisks denote the samples analyzed by transcriptomic sequencing in this report. Interfant-06 induction chemotherapy includes prednisone, dexamethasone, vincristine, cytarabine, daunorubicin, PEG-asparaginase, methotrexate, bortezomib and melphalan; AALL15P1 consolidation protocol includes cyclophosphamide, mercaptopurine, cytarabine, methotrexate, hydrocortisone and methotrexate. Dx = diagnosis; REF = refractory; REL = relapse; AZA = azacytidine; blina = blinatumomab; FLAG-IDA = fludarabine, cytarabine, granulocyte-colony stimulating factor, idarubicin.

Results
Transcriptomic sequencing (mRNA-Seq) performed on PBMNCs at diagnosis identified the KMT2A::AFF1 gene fusion, with low number of reads. Two IKZF1 gene fusions were also identified: IKZF1::TUT1 and KDM2A::IKZF1 ( Figure 2A, Table S2). Both IKZF1 fusions were validated by PCR and Sanger sequencing ( Figure 2B). Significantly, this is the first time the KDM2A::IKZF1 and IKZF1::TUT1 gene fusions have been described, and the first report of KMT2Ar ALL with co-occurring IKZF1 fusions [8]. Although KDM2A and TUT1 are both in the same karyotypic region of chromosome 11 (11q13.2 and 11q12.3, respectively), they are separated by >190 genes, with a genetic distance corresponding to a recombination frequency of >3.8% [9]. Thus, these two fusions likely represent separate genomic events. Multiplex ligation-dependent probe amplification (MLPA) is a PCR-based method for quantification of DNA copy numbers and a reliable method for copy number variation (CNV) genotyping. We used two different MLPA probe mixes (P202 and P335, MRC Holland) to determine CNV in genomic DNA. No deletions or duplications were detected in any of the genes assayed; of importance, no deletions were detected in IKZF1 exons 1-8 ( Figure S2).
The KMT2A::AFF1 gene fusion observed here is the most common fusion observed in infant disease (49% of all infant KMT2Ar leukemias [1]). However, the breakpoint in KMT2A exon 9 ( Figure 3A) is rarely observed in infant KMT2A::AFF1 leukemia (19%, compared with the frequently observed exon 11 breakpoint) [8]. Upon formation of KMT2A::AFF1, the entire C-terminal portion of KMT2A is lost; this region contains domains important for post-translational regulation and mediation of protein-protein interactions. The lost regulatory and H3K4 methyltransferase activity lead to the widespread epigenetic dysregulation observed in KMT2Ar patients [1]. The KMT2A portion retained in the fusion harbors binding motifs for proteins, such as menin and LEDGF, which are critical for leukemic transformation [10,11], and domains to facilitate KMT2A's DNA-binding capacity. The AFF1 portion of the fusion lacks a degron sequence, likely perturbing the protein's degradation rate. This is supported by high AFF1 expression in this patient, as compared with AFF1 expression in all other B-ALL samples in our patient cohort. Conversely, while KMT2A gene expression was not increased, elevated expression of the homeobox gene MEIS1 was observed as is typical of KMT2Ar patients [12] ( Figure 3B). It should be noted that, in an in vivo model, KMT2A::AFF1 fusions were incapable of inducing leukemia in isolation, suggesting additional genomic aberrations are required [13]. a recombination frequency of >3.8% [9]. Thus, these two fusions likely represent separate genomic events. Multiplex ligation-dependent probe amplification (MLPA) is a PCRbased method for quantification of DNA copy numbers and a reliable method for copy number variation (CNV) genotyping. We used two different MLPA probe mixes (P202 and P335, MRC Holland) to determine CNV in genomic DNA. No deletions or duplications were detected in any of the genes assayed; of importance, no deletions were detected in IKZF1 exons 1-8 ( Figure S2).  The KMT2A::AFF1 gene fusion observed here is the most common fusion observed in infant disease (49% of all infant KMT2Ar leukemias [1]). However, the breakpoint in KMT2A exon 9 ( Figure 3A) is rarely observed in infant KMT2A::AFF1 leukemia (19%, compared with the frequently observed exon 11 breakpoint) [8]. Upon formation of KMT2A::AFF1, the entire C-terminal portion of KMT2A is lost; this region contains domains important for post-translational regulation and mediation of protein-protein interactions. The lost regulatory and H3K4 methyltransferase activity lead to the widespread epigenetic dysregulation observed in KMT2Ar patients [1]. The KMT2A portion retained in the fusion harbors binding motifs for proteins, such as menin and LEDGF, which are critical for leukemic transformation [10,11], and domains to facilitate KMT2A's DNAbinding capacity. The AFF1 portion of the fusion lacks a degron sequence, likely perturbing the protein's degradation rate. This is supported by high AFF1 expression in this patient, as compared with AFF1 expression in all other B-ALL samples in our patient cohort. Conversely, while KMT2A gene expression was not increased, elevated expression of the homeobox gene MEIS1 was observed as is typical of KMT2Ar patients [12] ( Figure 3B). It should be noted that, in an in vivo model, KMT2A::AFF1 fusions were incapable of inducing leukemia in isolation, suggesting additional genomic aberrations are required [13].   [14]. (A) The KMT2A::AFF1 gene fusion retains DNA-and protein-binding domains, nuclear localization domains and the repression domain responsible for HDAC1 and HDAC2 recruitment. The menin-and LEDGF-binding domains are critical for leukemic transformation involving KMT2A, as both menin and LEDGF are essential oncogenic co-factors for KMT2A. The interaction of the three proteins is necessary for leukemic transformation [10,11]. (B) Boxplots of differential expression levels of five genes involved in gene fusions identified in patient CHI_0391. Gene expression levels in germline, diagnosis, refractory post-induction therapy and on-blinatumomab treatment samples from patient CHI_0391 (colored dots) are denoted and compared with samples from our B-ALL patient cohort (black dots; 592 patient samples at various disease timepoints including the 4 highlighted samples from CHI_0391. See Table S1 for cohort characteristics). The data are expressed as log normalized counts per million. The Terminal Uridylyl Transferase 1 (TUT1) gene has previously been reported as a fusion partner in T-cell lymphoblastic lymphoma patients [15]; however, IKZF1::TUT1 is a novel fusion gene. TUT1 encodes a nucleotidyl transferase enzyme that may play a role in controlling gene expression and cell proliferation; however, its role in oncogenesis remains unclear. The novel IKZF1::TUT1 fusion is the predominant gene fusion in this patient at all disease timepoints (Figure 2A, Table S2). Interestingly, only a small portion of the IKZF1 gene, containing no functional domains, is present. Thus, it is likely that the TUT1 fusion partner is driving the putative leukemic activity of this fusion. Additionally, the diagnosis sample exhibits the highest expression level of the TUT1 gene observed in our patient cohort ( Figure 3B). Limited functional data for TUT1 gene fusions exist; however, given that the RNA recognition motif is truncated while the nuclear localization sequence remains ( Figure 3C), it is possible that aberrant gene expression occurs as a result of IKZF1::TUT1.
The second novel fusion involving IKZF1, KDM2A::IKZF1, incorporates the catalytic domain of KDM2A and all IKZF1 functional domains ( Figure 3D). The encoded Ikaros protein is a transcription factor with key regulatory functions in lymphopoiesis [16]. IKZF1 is a leukemic driver and functions as a tumor suppressor and loss of Ikaros function, either by mutation or deletion, is frequently observed in B-cell ALL [17] as well as other hematological malignancies [18]. Ikaros loss of function alterations are associated with poor prognosis and inferior treatment outcomes [19][20][21]. However, currently no outcome data for IKZF1 gene fusions are available, most likely due to the rarity of these alterations (Table 1). Whether these fusions are driver alterations and contribute to leukemic development also remains to be determined.
Transcriptomic sequencing was performed on mesenchymal stem cells generated from hair follicles, representing a germline sample, as well as sequential samples taken when the patient was refractory following induction therapy and while undergoing blinatumomab therapy prior to relapse (Figure 2A). Interestingly, the KDM2A::IKZF1 and KMT2A::AFF1 gene fusions were no longer detectable by mRNA-Seq following blinatumomab therapy, suggesting the IKZF1::TUT1 gene fusion may be responsible for driving relapse.

Discussion
Here, we describe a rare case of congenital KMT2Ar ALL presenting with co-occurring IKZF1 gene fusions and a predictably aggressive disease trajectory. We report for the first time the novel IKZF1::TUT1 and KDM2A::IKZF1 gene fusions. Rearrangements involving KMT2A are commonly retained in relapsed infant ALL [5,31,32]; however, in this case, the KMT2A::AFF1 gene fusion did not appear to be the lesion driving leukemic relapse. Instead, our data suggest that relapse was driven by IKZF1::TUT1. This gene fusion remained in all samples investigated, including the on-blinatumomab therapy sample taken immediately prior to relapse. Conversely, the KMT2A::AFF1 gene fusion was only detected in the diagnosis and refractory post-induction samples, highlighting a key role for IKZF1::TUT1 in disease pathogenesis. Intriguingly, both IKZF1 gene fusions are predicted to be out-of-frame (Table S2); however, our data demonstrate the IKZF1 gene is still expressed ( Figure 3B). This is not unprecedented, and it has previously been observed that out-offrame fusions can cause transcriptional activation/repression of genes involved in the fusions leading to increases or decreases of their expression and the associated functional outcomes [29].
Ikaros is a lymphoid transcription factor with a tumor-suppressive function. Alterations that knock out the Ikaros function, such as the gene fusions described here, would presumably also affect Ikaros target genes including signal transducers (c-kit, Flt3, Il7r), pre-B-cell receptor signaling proteins (Syk) and cell cycle regulators (Cdkn2a, Cdkn1a) [33]. Indeed, altered Flt3 [34] and Syk [35] expression has been reported in KMT2Ar ALL. Studies have demonstrated Flt3 inhibitors are active against KMT2Ar disease in vivo [36] and, when administered in combination with various chemotherapeutics including some of those used here (dexamethasone, cytarabine, asparaginase), effectively kill KMT2Ar cells in vitro [37,38]. More recently, a Children's Oncology Group trial has demonstrated the benefit of Flt3 inhibitor lestaurtinib in combination with chemotherapy (Interfant 99-based induction regimen) for treating infants with KMT2Ar ALL [39]. Similarly, the combination of vincristine and the Syk inhibitor entospletinib demonstrated enhanced efficacy in in vivo models of infant KMT2Ar ALL compared with either agent alone [40]. However, entospletinib as a treatment for ALL has yet to enter clinical trials. A retrospective case study of 11 infants with KMT2Ar ALL demonstrated the efficacy of blinatumomab in patients with relapsed/refractory disease [41], and pre-clinical efficacy was observed in recent in vivo models assessing azacitidine in combination with venetoclax [42]. However, we observed poor response to the Interfant-06 induction and Children's Oncology Group consolidation protocols (AALL15P1), which comprise various chemotherapy agents, including those detailed above (Figure 1), and while azacitidine was well tolerated, the patient soon became resistant. The immunotherapies blinatumomab and inotuzomab [43] as well as the FLAG-IDA relapse regimen also failed. Case reports such as this highlight the urgent need for new targeted therapies to improve outcome in KMT2Ar infant ALL. We also demonstrate the importance of gene sequencing to comprehensively dissect the underlying genomic complexity of a disease like ALL and identify co-occurring alterations that may impact treatment outcomes.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/genes14020264/s1, Table S1: Key characteristics of the B-ALL patient cohort used in the gene expression analyses; Table S2: Complete range of gene fusions identified in patient CHI_0391 by transcriptomic sequencing; Figure S1: Immunophenotyping analyses of diagnosis, refractory and on-blinatumomab samples; Figure S2: Analysis of deletions in key B-ALL genes in CHI_0391 identified by multiplex ligand-dependent probe amplification (MLPA).  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in the study are deposited in the European Genome Phenome Archive.