The Biology, Pathogenesis and Clinical Aspects of Acute Lymphoblastic Leukemia in Children with Down Syndrome

Abstract

Children with Down syndrome (DS) are at a 20-fold increased risk for acute lymphoblastic leukemia (DS-ALL). Although the etiology of this higher risk of developing leukemia remains largely unclear, the recent identification of CRLF2 and JAK2 mutations and study of the effect of trisomy of Hmgn1 and Dyrk1a on B-cell development have shed significant new light on the disease process. Here we focus on the clinical features, biology, and genetics of ALL in children with DS. We review the unique characteristics of DS-ALL on both the clinical and molecular levels and discuss the differences in treatments and outcomes in ALL in children with DS compared to those without DS. The identification of new biological insights is expected to pave the way for novel targeted therapies.

Clinical Features and Treatment of DS-ALL

DS is one of the most common chromosomal disorders, occurring in approximately 1 in 800 births annually in the United States. DS is caused by trisomy for chromosome 21 (Hsa21) and is associated with several well established morbidities including cardiac defects, growth anomalies, endocrinopathies, ophthalmologic disorders and learning disabilities (1-5). Despite the fact that children with DS have a higher risk for malignant neoplasms (6), the incidence of solid tumors in people with DS is actually lower than in the population without DS, including all age groups (7, 8). However, there is clear predominance of leukemia including both acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL), which occur mostly in young patients. Moreover, children with DS are frequently born with transient myeloproliferative disorder, a pre-leukemia that is characterized by excessive growth of immature megakaryoblasts. In contrast to AMKL in children without DS, DS-AMKL blasts often co-express megakaryocytic and erythroid markers (9), perhaps owning to the unique presence of GATA1 mutations in the latter cells (10). Unlike AMKL in children and adults without DS, DS-AMKL is highly curable with standard reduced dose chemotherapy (11-13). This increased drug sensitivity is likely due to altered cytidine deaminase levels that are a consequence of the GATA1 mutations (14).

There are also some unique features of ALL in children with DS. These include the predominance of Caucasians (15), significant rarity of infant ALL (16, 17) and a paucity of T cell ALL (16-18). Moreover, favorable cytogenetics such as ETV6-RUNX1, double trisomy (trisomy of 4 and 10) and high hyperdiploidy, as well as unfavorable cytogenetics such as BCR-ABL1 and MLL rearrangements, are less common in DS-ALL than non-DS ALL (16-19). In the recent international Ponte di Legno study, which analyzed retrospective data from 16 clinical trials, the majority of the DS subgroup (40.3%) had a normal karyotype (i.e. other than the constitutional trisomy 21), which is in marked contrast to the low incidence in non-DS children (6.9%) (17). Standard therapy is generally used for DS-ALL, but the unique toxicity profiles of patients with DS bring challenges. A recent COG high-risk ALL study, AALL0232, which included prednisone or dexamethasone, vincristine, PEG-asparaginase and daunorubicin, and a standard-risk study, AALL0932, with dexamethasone, vincristine and PEG-asparaginase revealed excessive mortality during the induction phase in patients with DS (20). After the addition of extensive supportive care guidelines and leucovorin rescue for intrathecal methotrexate, mortality improved in the AALL0932 protocol, but not in AALL0232. Based on these studies, the 3-drug combination is used during induction of DS-ALL patients enrolled on the successor COG high-risk protocol AALL1131. In contrast, the Ponte di Legno study found no difference in treatment-related mortality between 3-drug induction and 4-drug induction, suggesting that there is no impact of anthracycline on treatment-related mortality (17). Another important observation of the Ponte di Legno study is that treatment-related mortality in DS was seen in all phases of treatment, including maintenance therapy, which rarely leads to death in ALL patients without DS (17).

Overall, the outcome of DS-ALL patients is worse than that of the general pediatric population. This difference is attributed to a higher relapse rate (17), which may be a consequence of dose reduction. For example, it is widely recognized that patients with DS have a higher incidence of methotrexate toxicity, which is most commonly gastrointestinal (21). Therefore, high-dose methotrexate is avoided or the dose is titrated in most protocols (17).

Table 1 summarizes the outcome of DS-ALL observed in studies published in the last decade. At least six different studies reported inferior outcome for the cohort with Down syndrome. For example, the Ponte di Legno study concluded that patients with DS-ALL had a higher 8-year cumulative incidence of relapse (26%) compared with the non-DS ALL reference cohort (15%) from the Dutch Child Oncology Group (DCOG) and the Berlin-Frankfurt-Muenster (BFM) study groups (17, 18). This study also found that age <6 years, white blood cell count <10 × 10⁹/L, and the presence of ETV6-RUNX1 were independent favorable prognostic factors for EFS, while age, ETV6-RUNX1 and high hyperdiploidy were associated with relapse-free survival. Furthermore, the study found a higher 2-year incidence of treatment-related mortality in the group with DS (7% versus 2% in patients with DS or non-DS, respectively) that was associated with a lower EFS. Separately, the Nordic Society of Pediatric Hematology and Oncology (NOPHO) ALL92 and the NOPHO ALL2000 protocols concurred that the 5-year EFS for patients with DS was poorer compared to those without DS (22). These authors concluded that the physician's lack of protocol adherence and decreased doses during maintenance phase may have contributed to the inferior outcome in DS patients. Indeed, the higher relapse rate may be a consequence of dose reduction.

Table 1. Summary of published studies on the outcome of DS-ALL.

Study (reference)	Population	Survival rates	Notes
International ALL Ponte di Legno Working Group (17)	4445 non-DS 653 DS 1995-2004	8-year EFS 64±2% DS vs. 81±2% non-DS (P<.001) 8-year OS 74±2% DS vs. 89±2% non-DS (P<.001) 2-year TRM 7±1% DS vs. 2±1% non-DS (P<.001)	B-precursor ALL only The DS cohort is from 16 international trials. Non-DS reference cohort is from the DCOG and the BFM study groups.
UKALL study (75)	3040 non-DS 86 DS 2003-2011	5-year EFS 65.6% DS vs. 87.7% non-DS (p<0.00005) 5-year OS 70% DS vs. 92.2% non-DS (P<0.00005) 5-year TRM 21.6% DS vs. 3.3% non-DS, p<0.00005)	The study reported 0/28 relapses for patients in the MRD low risk group.
NOPHO (76)	4637 non-DS 128 DS 1981-2010	5year-EFS(SE) 0.574 (0.057) DS vs. 0.783 (0.008) non-DS (P<.001) 5-year OS 0.691 (0.054) DS vs.0.894 (0.006) non-DS (P<.001)	B-precursor ALL only
COG (15)	2731 non-DS 80 DS 1999-2005	5-year EFS 69.9%±8.6% DS vs. 78.1%±1.2% non-DS (P=.078) 5-year OS 85.8%±6.5% DS vs. 90.0%±0.9% non-DS (P=.033)	B-precursor ALL only EFS and OS were similar in DS and non-DS when MLL rearrangement, BCR-ABL1, ETV6-RUNX1, trisomy of 4 and 10 were excluded.
AIEOP (18)	6237 non-DS 119 DS 1982-2004	10 year EFS (SE) 55.8% (4.9%) DS vs. 69.7% (0.7%) non-DS 10 year OS (SE) 60.4% (5.1%) DS vs. 79.7% (0.6%) non-DS	B-precursor ALL only
CCG (77)	8268 non-DS 179 DS 1983-1995	10 year EFS Standard risk 56.29±6% DS vs. 73.48±0.7% non-DS (P<.001) High-risk 62.06±6.5% DS vs. 59.29±0.9% non-DS P=.9 10 year OS Standard risk 70.17±4.9% DS vs. 84.87±0.6% non-DS (P<.001) High-risk 63.32±7.1% DS vs. 65.49±1% non-DS P=.7	Includes 8 DS and 747 non-DS with T-ALL

Abbreviations: DS; Down syndrome, SE; standard error, OS; overall survival, EFS; event-free survival, TRM; treatment-related mortality; AIEOP; Italian Association of Pediatric Hematology Oncology; NOPHO; Nordic Society of Pediatric Hematology Oncology; DCOG; Dutch Child Oncology Group; BFM; Berlin-Frankfurt-Muenster; COG, Children's Oncology Group

Although hematopoietic stem cell transplantation (HSCT) remains a viable option for those with high risk or relapsed ALL, the data on HSCT in patients with DS are limited and the role in relapsed patients with DS remains unclear (23, 24). New agents with novel mechanisms may be helpful in DS patients with refractory or relapsed disease, but unfortunately, patients with DS are excluded from many clinical trials for a concern of toxicity. There are several novel targeted therapies, which may help to improve survival in this population including chimeric antigen receptor (CAR) T cells which have shown significant activities in adult and pediatric ALL (25-29) and appear to be viable options for DS-ALL. Other immunomodulatory agents include blinatumomab, a CD19/CD3 bispecific antibody that redirects cytotoxic T-cells to CD19 expressing leukemic cells. This antibody, which has progressed through phase 1 and 2 studies and is currently in phase 3 in pediatric ALL (30), has been used in at least one patient with DS (31). At this point, it is unknown whether patients with DS will exhibit different clinical and biological responses to blinatumomab or CAR T therapy, although there is anecdotal evidence of success. Other potential new therapies include those that target JAK2 or mTOR, whose pathways are activated in the majority of DS-ALL cases (see “Genetics of DS-ALL”). Investigational agents include JAK inhibitors, such as ruxolitinib and momelotinib, or mTOR inhibitors including temsirolimus and everolimus (32-34).

Genetics of DS-ALL

Recent studies have provided key insights into other genetic contributions that play a role in the increased risk for and development of DS-ALL. These alterations include mutations in JAK2, NRAS and KRAS, mutations or overexpression of CRLF2, and trisomy 21 (Figure 1). While the specific genes on Hsa21 that participate in B-ALL have been unclear, recent studies implicate HMGN1 (35), and DYRK1A (36). Several other genetic events have been found in DS-ALL, including IKZF1 deletion (37, 38), PAX5 deletion (39, 40), ETV6-IGH rearrangement (41), and histone gene deletions (42). Given this extensive list of players involved in DS-ALL, it is not surprising that prior studies have emphasized the genetic heterogeneity of this subtype (43). The complexity of DS-ALL is underscored by the observation that development of an animal model of DS-ALL required the combination of a Jak2 mutation, CRLF2 overexpression, Pax5 deficiency, Ikzf (Ikaros) loss and the mouse equivalent of trisomy 21 (35). In the following sections, we review the major genetic changes involved in DS-ALL and how these affect disease progression.

Figure 1. Schematic of genetic events implicated in the etiology of DS-ALL.

Several genetic events cooperate in the pathogenesis of DS-ALL, with trisomy 21 being an underlying factor. While overexpression of CRLF2 (A) or JAK2 activating mutations (B) alone are not sufficient to induce cytokine-independent growth, the combination of the two genetic aberrations can induce constitutive JAK-STAT signaling and promote leukemia (C). Recent evidence has indicated that NRAS and KRAS mutations (D) occur in some cases and are mutually exclusive with JAK2 mutations, indicating that other signaling components can promote proliferation downstream of CRLF2 mutations and other inciting events. Similarly, it has also been found that gain-of-function IL7R mutations (E) allow for dimerization with CRLF2 and cytokine-independent growth, which is further enhanced in the presence of TSLP (denoted by bolded line). However, despite of the prevalence of these driver mutations, many cases of DS-ALL occur without this combination of events. Recent evidence suggests contributions of trisomy of HMGN1 and DYRK1A and deletions of PAX5 and IKZF (F). Of note, the latter two events were needed in combination with CRLF2 overexpression and a JAK2 activating mutation for development of an animal model of DS-ALL (35). These genetic alterations exhibit an enormous variability in both nature and frequency, underscoring the complexity behind the pathogenesis of DS-ALL.

Driving Events in Ds-All

Alterations in CRLF2

Among the genetic abnormalities implicated in development of DS-ALL, aberrant expression of cytokine receptor like factor 2 (CRLF2) has been well characterized. CRLF2 is an important lymphoid signaling receptor, which complexes with IL-7Rα to form a heterodimeric receptor for thymic stromal lymphopoietin (TSLP). Several groups discovered that CRLF2 is overexpressed or mutated in a subset of B-ALL cases that lack common translocations such as t(12;21) and t(1;19), particularly DS-ALL (43-46). Overexpression of CRLF2 is caused by both novel deletions of the pseudoautosomal region (PAR1) of Xp22.3/Yp11.3 and by immunoglobulin heavy chain translocations. Specifically, the PAR1 deletion was shown to be adjacent to the CSF2RA and CRFL2 genes. Array-based comparative genomic hybridization uncovered that the PAR1 deletion resulted in a rearrangement between first non-coding exon of P2RY8 and the entire coding region of CRLF2 (44). P2RY8 encodes a purinergic receptor that is expressed highly in a variety of tissues; indeed, this fusion results in higher CRFL2 expression (47). Moreover, CRLF2 overexpression has also been seen in B-ALL cases due to IGH@-CRLF2 fusion (43, 45). These rearrangements are fairly rare in non-DS-ALL, but common in DS-ALL with rates of approximately 5% and 50%, respectively. Of note, a point mutation in CRLF2 resulting in F232C was also identified in approximately 9% of DS-ALL patients and 21% of adult B-ALL patients who overexpress CRLF2 (43, 46). It is also noteworthy that the CRLF2 abnormality was seen in nearly 50% of cases of Ph-like ALL, with more than half of these also harboring JAK1/JAK2 mutations (48). Overall, CRLF2 dysregulation is found in 5-10% of pediatric ALL and in 60% of DS-ALL, indicating its significant role as an inciting factor in the development of the DS subtype (49).

JAK2 mutations

Mutations in JAK2 are commonly seen in hematopoietic disorders; V617F mutations are very common in patients with myeloproliferative neoplasms (MPNs), accounting for more than 90% of polycythemia vera and nearly 50% of essential thrombocythemia and primary myelofibrosis (50). As JAK-STAT signaling also plays an important role in B lymphopoiesis, several studies have not surprisingly revealed a role for JAK2 mutations in the etiology of acute lymphoblastic leukemia, particularly DS-ALL. The first case of a JAK2 activating mutation in a patient with DS with pre-B ALL was described in 2007 (51). The report revealed the presence of a novel deletion of five amino acids (JAK2∆IREED) within the JH2 pseudokinase domain from a screen of JAK2 in 90 cases of acute leukemia of myeloid or B cell origin. The authors demonstrated that expression of this deletion mutant conferred constitutive JAK-STAT signaling to Ba/F3 cells similar to that seen upon V617F expression, a hallmark of primary myelofibrosis. As the only other genetic event present was trisomy 21, the patient observations suggest a cooperative relationship between trisomy 21 and JAK2 activation. Next, a 2008 study found that 16 of 88 patients with DS-ALL harbored R683 mutations in bone marrow specimens, while none of the other 72 samples presented any other JAK2 exon 16 mutations (52). Of the missense mutations found in these 16 patients, 8 were R683G, 5 were R683S, 1 was R683K, and 2 were large insertions before R683. It was further revealed in this study that expression of JAK2 mutants in BaF3/EpoR and BaF3/TpoR cells resulted in cytokine-independent proliferation and constitutive phosphorylation of JAK2 and STAT5. A second study reported in an independent cohort of pediatric DS-ALL patients that 10 of the 53 subjects carried a point mutation in R683, the most prevalent being R683G (53). Moreover, through analysis of paired germline DNA samples, available for 5 of the DS-ALL patients with R683 mutations, the authors concluded that all of the mutations were somatic events. A third study identified mutations in JAK1, JAK2, and JAK3 in children with high-risk B-ALL at a rate of 20% (54), though beyond the context of DS these mutations are generally rare, with the exception of Ph-like ALL (48). Finally, another study identified JAK2 R683 mutations in 28% of DS-ALL cases (40). Ultimately, JAK2 mutations have been identified in approximately 21% of DS-ALL patients, the majority of which are CRLF2-positive.

Why are V617F and R683 mutations restricted to the MPNs and B-ALL, respectively? The R683 and V617 residues both lie in the JAK2 pseudokinase domain, but on different highly conserved sites (55). The DS-ALL mutations are predicted to affect the interactions with the C-terminal kinase domain, resulting in constitutive JAK2 activation, similar to the V617F mutation in the MPNs. Moreover, it is thought that both V617 and R683 mutations disrupt the interaction of the JAK2 pseudokinase domain with the JH1/JH2 domains, thereby reducing the intrinsic negative regulatory activity. However, the mechanism behind the selectivity of these mutations in AML versus ALL is still poorly understood. One report has described two structural hotspots in the JAK2 pseudokinase domain containing either V617 (hotspot I) or R683 (hotspot II). The authors postulate that mutations in these hotspots, in addition to activating JAK2, may alter the recruitment to different signaling complexes, thereby providing a potential mechanism behind the genotype-phenotype specificity of JAK2 pseudokinase mutations (56).

Cooperation between JAK2 and CRLF2

Several studies have reported an association between aberrant CRLF2 expression and JAK2 mutations in both B-ALL and DS-ALL (43-46). While JAK2 mutations are most commonly found in the R683 pseudokinase domain, mutations have also been observed in the JAK2 kinase domain (T875N and G861W) and the JAK1 pseudokinase domain (V658F, the corresponding to the V617 residue of JAK2)(44). In one study of a validation cohort of 53 DS-ALL cases, 20 harbored the PAR1 deletion resulting in CRLF2 rearrangement; 8 of the 53 patients also harbored a JAK2 mutation, whereas the JAK2 mutation was only observed once in subjects without the PAR1 deletion (44). A different study from the International BFM Study Group found that 10 out of the 10 DS-ALL samples with JAK2 R683 mutations also exhibited CRLF2 overexpression (43). These data suggest a tight association between the occurrence of CRLF2 and JAK2 genetic alterations in DS-ALL etiology.

This high incidence of co-occurrence of the two events suggests that there is cooperation between CRLF2 and JAK2 alterations. Indeed, while expression of P2RY8-CRLF2 or JAK2 mutants alone in Ba/F3-IL7R cells was insufficient to induce cytokine-independent growth, expression of both led to constitutive JAK-STAT signaling and cytokine independence (44). Similar experiments were performed in BaF/3 cells that express CRLF2 by co-expression of wild type JAK2 or the R683S mutant. Although JAK2 R683S expression allowed for cytokine independent growth, the effect was significantly amplified when R683S was co-expressed with CRLF2 (43). The F232C CRLF2 mutation similarly cooperates with JAK2 mutations in promoting cytokine independent growth (43, 46). However, despite the prevalence of CRLF2 overexpression and JAK2 activating mutations in DS-ALL, many still lack these well-characterized lesions, implying that there are other components of the disease that still require further study, several of which are described below.

In addition to JAK mutations driving JAK/STAT signaling in malignant B cells, a recent study identified mutations in the IL7 receptor in patients with B- and T-ALL (57). In B-ALL the mutations were associated with aberrant CRLF2 activation, forming a receptor for TSLP. The IL7R mutations included a S185C mutation in the extracellular domain or, more commonly, an in-frame indel in the transmembrane domain. These mutations were found to confer constitutive activation of IL7R and allow for cytokine-independent growth in conjunction with CRLF2 overexpression. This effect was augmented by TSLP.

Not all cases of CRLF abnormalities are associated with JAK2 mutations. One recent study reported the identification of mutations in NRAS or KRAS in 15 of 42 DS-ALL cases, which is similar to the incidence of JAK2 or P2RY8-CRLF2 fusions (58). The RAS mutations were mutually exclusive with JAK2 mutations. Interestingly, it was found that JAK2-mutated or PTPN11-mutated sub-clones had switched to a RAS-mutant clone in two of three relapsed cases. The interplay between these multiple signaling mechanisms highlights both the complexity and the importance of the CRLF2-JAK2 axis in promoting leukemic potential.

Chromosome 21 contributions to DS-ALL

Multiple mouse models of DS, including the Ts65Dn and Ts1Rhr strains that are trisomic for 104 and 31 orthologs of Hsa21 genes respectively, develop a progressive myeloproliferative disorder characterized by increased megakaryopoiesis and thrombocytosis (59, 60). Moreover, combining the partial trisomy of the Ts1Rhr mice with a Gata1 mutation and an activating mutation in MPL leads to an aggressive DS-AMKL like phenotype in vivo (60). Recently, the validity of using mouse models to study leukemia in humans with DS was extended by a demonstration that mouse models of DS also develop B-ALL (35). The disease required not only a JAK2 mutation and CRLF2 overexpression, but also heterozygous loss of Pax5 and expression of a dominant negative Ikzf isoform in the Ts1Rhr bone marrow. These findings reveal DS-ALL is indeed a complex genetic disorder.

While trisomy 21 is clearly associated with leukemia, the specific gene(s) on this chromosome that promotes B-ALL when present in three copies remains unclear. Two recent studies have implicated specific Hsa21 genes, HMGN1 and DYRK1A, in lymphopoiesis and the disease process (35, 36).

The nucleosome remodeling protein HMGN1

To uncover how trisomy 21 promotes lymphoid neoplasia, Weinstock and colleagues carefully analyzed lymphopoiesis in the Ts1Rhr mouse model. First, they compared the ratio of B cells that fell into the different Hardy fractions, which were described in 1991 as a way to classify B cell precursors based on the expression of cell surface proteins (61). They discovered that there was an accumulation of B cells in the less mature Hardy A fraction with concomitant reductions in the more mature B and C fractions in the Ts1Rhr mice (35). There was also an overall reduction in B cell numbers. Next, they investigated the colony forming activity of B cell progenitors and made the surprising observation that pre-B cells from the trisomic animals had the ability to re-plate for at least 10 generations, despite the reduction in their overall numbers. This re-plating phenotype indicates that the presence of three alleles of one (or more) of the 31 trisomic genes confers progenitor B cell self-renewal. Finally, they introduced the BCR-ABL oncogene into disomic or Ts1Rhr bone marrow and transplanted the cells to irradiated recipients. BCR-ABL expression in the Ts1Rhr genotype gave rise to a disease with shorter latency and increased penetrance than in euploid controls.

Several lines of evidence suggest that Hmgn1 is one of the critical trisomic genes that alters B-cell development in Ts1Rhr mice. First, shRNAs against Hmgn1 were selectively depleted during Ts1Rhr pre-B cell colony replating (35). This argues that persistent expression of the gene is needed for self-renewal. Second, overexpression of HMGN1 in Ba/F3 cells suppressed H3K27me3 levels. This is consistent with the global reduction in H3K27me3 observed in the trisomic pre-B colonies. Third, overexpression of HMGN1 alone was sufficient to confer a more aggressive BCR-ABL disease on euploid progenitors. Together, this study points to HMGN1 as a critical B-cell leukemia-promoting gene on Hsa21. The study also points to a new possible avenue for therapy: GSK-J4, a small molecule inhibitor of H3K27 demethylases, impaired Ts1Rhr B-cell colony formation and re-plating activity. Future research with GSK-J4 and analogs with better bioavailability will shed more light on the potential for targeting H3K27 methylation in this disease.

Interestingly, it has been reported that trisomy 21 in human fetal liver is associated with impaired B cell development as evidenced by a reduction in committed B-lineage progenitors, as well as a decrease in the ability of HSCs, lymphoid-primed multipotent progenitors (LMPPs) and early lymphoid progenitors (ELPs) to develop into mature B cells (62). This reduction in pre-proB cells and pro-B cells parallels the reductions in B cell fractions in the Ts1Rhr animals. However, the extent to which the in vitro expansion of Ts1Rhr pre-B cells recapitulates pre-leukemia events in humans remains to be determined.

The dual specificity kinase DYRK1A

One of the major players in the neuronal pathogenesis of DS is DYRK1A (dual-specificity tyrosine phosphorylation-regulated kinase 1A), transcribed from the DSCR of chromosome 21 (63). A member of the CMGC superfamily, DYRK1A has been well characterized in various tissues, particularly neurological models, with roles including cell proliferation and differentiation (64), RNA splicing (65), and apoptosis (66). DYRK1A was not studied in hematopoiesis until an AMKL-promoting role for the kinase was discovered by virtue of its dysregulation of nuclear factor of activated T cells (NFAT) activity (60). While this inhibition of NFAT signaling by DYRK1A has canonically been linked to lower incidence of cancers in adults with DS (67), the data in this former study suggests that increased NFAT dysregulation promotes AMKL in children with DS, thereby providing evidence for DYRK1A having both cancer-promoting and cancer-inhibitory activities.

DYRK1A may play a role in promotion of ALL, but, likely through a pathway different from that in AMKL, as Dyrk1a was recently shown to be essential for lymphoid, but not myeloid, cell development (36). Ablation of Dyrk1a in the proliferative stages of pre-B and pre-T cells led to a transcriptional up-regulation of cell cycle-promoting genes. However, while these cells appeared to be in the S-G2-M cycling state, there was a decrease in total cell number and an inability of pre-B colonies to proliferate ex vivo. There have been previous studies suggesting a significant role for DYRK1A in cell cycle regulation in other tissues, such as phosphorylation of Lin52 (S28) and consequential assembly of the cell cycle regulatory DREAM complex (68) or phosphorylation and stabilization of p27 (69). Within developing lymphocytes, DYRK1A phosphorylates cyclin D3 on T283, a highly conserved phosphodegron site among D-type cyclins (36). Furthermore, degradation of phosphorylated cyclin D3 was associated with repression of E2F target genes. This finding complements a previous study, which showed that DYRK1A phosphorylation of cyclin D1 on the homologous T286 induced G₀ cell cycle exit and differentiation of fibroblasts (70). Thus by negatively regulating cyclin D3 stability, DYRK1A plays a major role shifting lymphocytes from proliferation to quiescence. Despite the fact that Dyrk1a deficient cells fail to enter quiescence, they did not accumulate in the animal (36). Rather, large pre-B cells lacking Dyrk1a completed fewer cell divisions than control cells, indicating that DYRK1A has additional roles in cell growth. The requirement for Dyrk1a in lymphopoiesis suggests that it may be a novel target for therapeutic intervention. Importantly, loss of Dyrk1a had no effect on the myeloid lineages, further supporting the potential of DYRK1A inhibition as a targeted therapy for ALL.

Other Genetic Alterations

In addition to the previously discussed genetic alterations associated with DS-ALL, there are several other genetic events that have been observed. Among these, deletions of IKZF1 and PAX5 are often cited, both of which are prevalent in non-DS-ALL (39, 71). While loss of IKZF1 alone does not lead to development of B-ALL, in the context of BCR-ABL1, IKZF1 haploinsufficiency reduced the latency in development of B-ALL in transgenic mice (37). Similarly, deletion of IKZF1 has been observed in approximated 75% of BCR-ABL1 positive- BALL cases. This finding suggests that while loss of IKZF1 is not a primary event in leukemic development, it may be an essential secondary event. In murine models, it has been shown that the dominant-negative isoform tends to have a more severe leukemic phenotype, although haploinsufficiency accelerates the development of leukemia (37). Thus despite the different effects of these deletions, both cases are strongly implicated in hematologic neoplasia. In DS-ALL patients, the frequency of IKZF1 deletions (26-35%) is comparable to high-risk non-DS-ALL patients (29%) based on independent cohorts from the Dutch Childhood Oncology Group (DCOG) and UK trials (39). Interestingly, compared to B-ALL patients, DS-ALL patients have a significantly lower incidence of combined IKZF1 deletion and JAK2 mutations, which provides further evidence that genetic patterns causative of DS-ALL differ significantly from non-DS-ALL. PAX5 is another gene involved in B-cell development that has been implicated in DS-ALL. Analysis of the DCOG cohorts revealed deletions in PAX5 in 12% of DS-ALL patients, and the event was not mutually exclusive to IKZF1 deletions (39). Though this frequency is less than what is observed in B-ALL patients, it nonetheless underscores the recurring theme of varying degrees of similarity between the contributing factors in non-DS B-ALL and DS-ALL.

iAMP21

Another causative genetic event in ALL that is related to DS-ALL is an intrachromosomal amplification of chromosome 21 (iAMP21), which has been reported in 2% of pediatric ALL (72). iAMP21 rarely arises from a germline Robertsonian translocation, rob(15;21)(q10,q10)c, or more commonly is sporadically initiated by breakage-fusion-bridge (BFB) repair cycles. Both of these events display evidence of chromothripsis as an inciting mechanism (73) and may result in duplications of the abnormal chromosome, providing an optimal gene dosage for leukemogenesis. While rob(15;21)c only accounts for 0.5-1% of all Robertsonian translocations, patients with this chromosomal abnormality have a ∼2700-fold increased risk in developing iAMP21-ALL (73). Genetic analysis of patients with iAMP21 has revealed several interesting patterns. For example, one study showed that while RUNX1 is often found in the amplified region of patients with iAMP21, and RUNX1 translocations with ETV6 are commonly associated with ALL, RUNX1 mutations are generally not seen in iAMP21 (74). Similarly JAK mutations were seen in none of the 15 iAMP ALL patients in that study. Additionally, of 94 samples from iAMP21 patients, the group observed IKZF deletions (16%), PAX5 deletions (8%), CDKN2A deletions (13%), ETV6 deletions (15%), RB1 deletions (13%), gain of X chromosome (20%), and P2RY8-CRLF2 fusion (17%). Moreover, although iAMP21 has always been characterized in patients with non-DS-ALL, many of the genes amplified are in the DSCR of chromosome 21, such as RUNX1 and miR-802. Another study that analyzed a minimal iAMP region revealed that that the most highly amplified regions of chromosome 21 includes RUNX1, DYRK1A and ETS2 (73). Certain genetic elements that are characteristic of DS-ALL are also seen in similar frequency in iAMP21 ALL, most notably gain of X chromosome (20% in iAMP21 ALL, 24% in DS-ALL) and P2RY8-CRLF2 fusion (17% in iAMP21 ALL, 22% in DS-ALL) (74). Nevertheless, there are still glaring differences, such as the lack of JAK mutations in iAMP21, despite the high rate of occurrence in DS-ALL. Thus the differences between iAMP21 ALL and DS-ALL underscore the complexity of genetic events involved in B-cell leukemia.

Conclusions and Future Directions

DS-ALL is a unique entity that is characterized by a largely distinct subset of mutations, including trisomy 21, high rates of CRLF2 overexpression, and JAK2 mutations. Although the overall responses are similar to children without DS, patients with DS-ALL are at higher risk for relapse and for severe side effects as a consequence of their hypersensitivity to conventional chemotherapy. In contrast to DS-AMKL, which has a well-defined genetic basis of GATA1 mutations on the X chromosome (10), the pathogenesis of DS-ALL is far more complex. Development of an animal model of DS-ALL required four events beyond trisomy: CRLF2 overexpression, a Jak2 mutation, and losses of Pax5 and Ikzf (35).

There are many questions to consider as we move towards novel therapies for DS-ALL. First, what are the specific genes that contribute to the leukemia predisposition? The study by Lane et al compellingly demonstrated that HMNG1 is a strong candidate pro-leukemia gene, while the study by Thompson et al provided evidence that DYRK1A is at least required for B-ALL. There are 29 other candidate genes in the Ts1Rhr model and well over 100 others from Hsa21 that need to be considered. With the advent of CRISPR technology, it is likely that experiments to identify the genes in an unbiased fashion will more quickly forward. Second, why is there a strong link between JAK2 mutations and CRLF2 alterations? Third, why are there different categories of JAK2 mutations in ALL versus the MPNs? Finally, what does the future hold with respect to novel and more effective therapies for DS-ALL? Studies to test JAK inhibitors, CAR-T cells, and immunomodulatory agents that are ongoing or planned will certainly advance the field. Given the multitude of health issues that individuals with DS face, it is imperative that we move aggressively forward to find less toxic and more efficacious treatments.