Review

The paradox of Myeloid Leukemia associated with Down syndrome

Images of individuals with physical characteristics of Down syndrome have been depicted in ancient artwork since antiquity. Amongst these, are images recovered from the Tolteca culture in Mexico (500 CE) and in European Renaissance paintings dating back to the 15th-17th century [3]. The term Down syndrome was named after the British physician John Langdon Down who first described in the medical literature, the physical appearance of cognitively impaired individuals as “Mongolism” in 1866 [4]. In 1959, the French geneticist Professor Jerome Lejeune studied the chromosomal karyotype of three men who had the distinct phenotype of Down syndrome and identified that the underlying genetic basis, was the presence of an extra copy of chromosome 21 (Trisomy 21) [5]. This finding was confirmed by Jacobs et al and Book et al who analyzed cells from other individuals with Down syndrome [6], [7].

Brewster and Cannon were the first to report a Down syndrome child having acute leukemia in the 1930s [8]. It is now recognized that children with Down syndrome have a greater predisposition to develop both acute lymphoblastic leukemia (ALL) and AML compared to children without Down syndrome [9]. The risk of developing ALL and AML is estimated to be 33 and 150 times greater, respectively, in patients with Down syndrome than the general population [10].

The 2008 World Health Organization (WHO) classification was the first to recognize myeloid proliferation in Down syndrome as a separate entity [11]. The 2016 WHO classification maintained this subgroup which consisted of a spectrum ranging from transient abnormal myelopoiesis (TAM) to myeloid leukemia associated with Down syndrome (ML-DS) [12].

Approximately 30% of newborns with Down syndrome develop a clonal neonatal preleukemic syndrome called TAM [9]. In TAM, blasts are detectable in the peripheral blood with the same morphology and immunophenotype as typically observed in acute megakaryocytic leukemia (AMKL). The peripheral white blood cell count and blast percentage can vary widely with some patients having WBC counts > 100,000/µL and extensive infiltration into organs including the liver.

Remarkably, the blasts can spontaneously regress in a large proportion of the TAM cases with normalization of blood counts with supportive care alone (e.g., transfusions). However, chemotherapy may be beneficial for certain subgroup of neonates with progressive life-threatening symptoms including hydrops fetalis, extreme leukocytosis (WBC > 100,000/µL), hepatic dysfunction, disseminated intravascular coagulation, renal and/or cardiac failure, as the mortality rate in these situations may be up to 20%. [13], [14]. Muramatsu et al [15] studied the risk factors contributing to early death (<6 months) in children with TAM. TAM patients were stratified based on their peak WBC count and gestational age with high-risk patients receiving low dose Ara-C. They identified that early gestational age (<38 weeks of gestation) was associated with early death due to cardiopulmonary, hepatic or renal failure. A possible explanation for early death was that in preterm infants, the immaturity of organs caused inability to tolerate blast infiltration, which led to organ failure in utero and hence preterm delivery.

TAM blasts have been shown to be very sensitive to Ara-C therapy, even at very low doses [16], [17]. The Ara-C treatment dose recommended by Berlin-Frankfurt-Münster (BFM) group is 0.5–1.5 mg/kg for 3–12 days for neonates with TAM and thrombocytopenia, hepatic dysfunction, and/or leukocytosis (>50,000/µL) [18]. Using low dose Ara-C in the high-risk subgroup of TAM patients listed above appeared to have a reduced early death rate as suggested in the TAM-10 study [19]. The Children’s Oncology Group (COG) Study A2971 enrolled 135 Down syndrome patients diagnosed with TAM, 38 of them had life-threatening symptoms, of which, 24 received Ara-C given at a dose of 3.33 mg/kg over 24 h for 5 days and repeated based on clinical status and resolution of compromising signs and symptoms [14]. However, likely due to the higher Ara-C dose and/or administration via a continuous infusion, there was a high rate of severe myelosuppression besides disease severity contributing to a low survival rate of only 51% in the group who were treated [14], [20].

Approximately 20% of newborns, who have resolution of TAM, will ultimately develop ML-DS. [21], [22] There is no evidence that the treatment of life-threatening TAM with Ara-C can prevent its progression to ML-DS [14], [20].

Roberts et al [23] prospectively analyzed the characteristics of 200 Down syndrome neonates including peripheral blood smear review and screening for the presence of somatic GATA1 mutations (described in more detail below). Peripheral blood blasts were observed in 97.5% of cases and GATA1 mutations were detected in 8.5% using Sanger sequencing/denaturing high performance liquid chromatography (Ss/DHPLC). Surprisingly, mutations were detected in a higher proportion of the cases (18 patients: 20.4%) using targeted next-generation sequencing (NGS). However, no clinical or laboratory features of TAM were noted in these 18 patients. Hence, they were labeled as “silent TAM”. This suggests that GATA1 mutation testing besides blood count and peripheral blood smear review is important to identify all Down syndrome babies at risk of ML-DS. Ss/DPHLC can be used for initial screening and NGS can be used to identify neonates with low levels of GATA1 mutant clones [23].

ML-DS typically is diagnosed before 3 years of age (mean age of 21 months) and it is extremely rare to be diagnosed after 4 years of age [24]. A three hit model of leukemogenesis in Down syndrome has been proposed by Mateos et al. [25] (Fig. 1).

Leukemogenesis resulting in TAM/ML-DS begins late in the first trimester in fetal development. Trisomy 21 causes perturbed hematopoiesis in fetal liver hematopoietic stem and progenitor cells resulting in an increase in the megakaryocyte-erythrocyte progenitors [20], [26].

Second hit: Prenatally acquired GATA1 mutations in the presence of Trisomy 21

The hematopoietic transcription factor gene, GATA1 (localized to Xp11.23), is essential for megakaryocyte, erythrocyte, mast cell and eosinophil development [27], [28]. The importance of somatic GATA1 mutations in ML-DS was first identified by John Crispino’s laboratory [29].

Somatic mutations (typically in exon 2) of the transcription factor gene GATA1 produces a truncated protein called GATA1s (GATA1s, a 40-kDa protein compared to the 50-kDa wild-type protein) which has reduced transactivation activity compared to the full-length protein, causing disruption in development of megakaryocytic progenitors [30], [31], [32] (discussed in more details in the GATA1 mechanism section). The combination of GATA1 mutations and Trisomy 21 predisposes to TAM.

For the most part, a low peripheral blood blast population (i.e. ≤ 10%) with GATA1 mutations leads to silent TAM and typically no clinical implications in the newborn period. On the other hand, >10% blasts may cause clinical TAM in the neonatal period [23]. A larger proportion of this clonal population becomes extinct over time. But approximately 10–20% of this larger clonal population is subjected to the acquisition of various additional genetic and epigenetic alterations [20]. Apart from the hallmark GATA1 mutations in ML-DS, other genetic alterations in TAM clones which lead to the progression to ML-DS include mutations in JAK1, JAK2, JAK3, TP53, FLT3 and MPL genes [33], [34], [35], [36]. Other mutations in cohesin genes (53%) (RAD21, STAG2, SMC3, SMC1A), CTCF (20%), epigenetic regulators (45%) (EZH2, KANSL1) and RAS pathway genes (NRAS, KRAS, CBL, PTPN11, NF1) are also known to contribute to the final hit and lead to the development of ML-DS [37], [38].

Wagenblast et al [39] studied the cellular mechanisms that lead to leukemic transformation. Their study established that Trisomy 21 is a pre-requisite for the initiation of GATA1s-induced preleukemic clones in hematopoietic stem cells. This is achieved by up-regulation of three chromosome 21 localized microRNAs, miR-99a, miR-125b-2 and miR-155, within the fetal hematopoietic stem cells. Once the pre-leukemic state is established, further transformation to leukemia is independent of Trisomy 21 (which was first discovered in this study) and is in fact dependent on deletion of cohesin genes such as STAG2. The CD117+/KIT proto-oncogene was also identified as the important driver for leukemia evolution and propagation to ML-DS. Pharmacologic KIT inhibition aimed at the preleukemic stage, could be a very important future approach to prevent transformation of preleukemic TAM to full blown ML-DS [39]. Another microRNA, mi-486-5p, which cooperates with GATA1s, was found to be overexpressed in ML-DS cells, leading to growth, survival and proliferation of ML-DS cells [40].

The GATA family is composed of a group of transcription factors; each of them has two zinc finger domains: a C-terminal zinc finger that is required for recognition and binding of the (A/T) GATA (A/G) consensus sequence and an N-terminal zinc finger that plays a role in stabilization and recruitment of cofactors [41]. The N-terminal transactivation domain is the source of variations among the GATA family proteins [41], [42]. GATA1 is considered the founding member of the GATA family, and it was first identified as a protein binding the 3′ enhancer of the beta-globin gene and was cloned from mouse erythroid linage in 1989 [43]. GATA1 is known to be expressed in megakaryocytes, erythroid cells, mast cells, eosinophils, basophils, and Sertoli cells of the testis [44].

The GATA1 gene comprises six exons. The GATA1 pre-mRNA could undergo alternative splicing producing two mRNA transcripts. One of the transcripts contains exons 1–6 which can be translated into two protein isoforms, a full-length 413 amino acid protein (GATA1) and a truncated short 330 amino acid isoform (GATA1s). The second transcript is missing exon 2 but retains exon 1 and exons 3–6, and this is only translated into the GATA1s isoform. The two zinc finger domains are present in both GATA1 and GATA1s; however, only GATA1 has the transactivation domain [44], [45], [46]. Therefore, GATA1s has similar DNA binding activity, but reduced transactivation ability, in comparison to the full-length protein.

Several studies have extensively demonstrated the importance of GATA1 in the development and maturation of erythroid and megakaryocytes, besides its role in eosinophil, mast and dendritic cell cycles [27], [28], [44], [47].

Several GATA1 mutations have been observed in different myeloproliferative disorders, uniquely noted in Down syndrome and not commonly in non-Down syndrome childhood leukemias. Mutations to the region of the GATA1 gene which encodes for the N-terminal transactivation domain were observed. These mutations hinder the translation initiation from the first AUG codon in exon 2, leading to translation initiation only from the alternative AUG codon in exon 3, resulting in exclusive synthesis of GATA1s isoform. [46], [48] These mutations were detected in ML-DS [29] and TAM [49], [50].

It is well established that GATA1 mutations are acquired prenatally, supported by the finding of identical twin Down syndrome infants with TAM and the identical GATA1 exon 2 mutations [51]. A retrospective study done in the past reported that GATA1 mutations were detected in Guthrie newborn screening cards in those infants with Down syndrome who later went on to develop ML-DS [52]. Moreover, a strong evidence supporting this fact is the detection of GATA1 mutations in fetal liver samples of patients with Down syndrome seen as early as twenty-one weeks of gestation [53].

GATA1 mutations are found specifically in Down syndrome patients (including mosaicism of chromosome 21). This brings up a very important question: Why would an X-linked chromosomal mutation (GATA1) be specific to chromosome 21? Although there is no evidence to support any association between the two chromosomes, a possible explanation is that trisomy 21 induces a “mutator phenotype”.

Down syndrome cells are prone to DNA repair defects. In these cells, baseline DNA repair is decreased, and they are increasingly sensitive to phytohemagglutinin, N-methyl-N‘-nitro-N-nitrosogluanidine and gamma-radiation. Due to this heightened sensitivity they are at an increased susceptibility for DNA oxidation, methylation, and DNA strand breaks [54]. Analysis performed by Cabelof et al [55] to study the effects cystathionine-β-synthase (CBS) and superoxide dismutase (SOD) on generation of GATA1 mutations in Down syndrome, identified several mutations at G:C > T:A, G:C > A:T and A:T > G:C including small insertion/deletion, duplication, and base substitution. This finding suggested that overexpression of SOD and CBS had an essential role in oxidative stress and aberrant folate metabolism leading to leukemogenesis secondary to GATA1 acquisition. Another mechanism in which mutations may arise in Down syndrome is increased uracil incorporation into DNA as a result of functional folate deficiency associated with CBS overexpression [56]. Moreover, loss of β-pol (DNA polymerase beta), the rate-limiting enzyme in the BER pathway, will potentially result in higher susceptibility to the mutagenic effects of unrepaired endogenous damage due to high levels of uracil incorporation [55].