Mature Analysis of the Prospective Oxford Down Syndrome (DS) Cohort Study: Timing and Clinical Impact of Preleukemic GATA1 Mutations and Lessons for Management of Newborns with DS

Children with Down syndrome (DS) have a high risk of developing a unique form of AML known as myeloid leukaemia of DS (ML-DS) defined by the presence of one or more acquired N-terminal truncating mutations in the hematopoietic transcription factor gene GATA1 . ML-DS is confined to children <4 y of age and preceded by a neonatal preleukemic syndrome, Transient Abnormal Myelopoiesis (TAM). Although the same GATA1 mutations are present in ML-DS and TAM, indicating they are clonally-linked disorders, chemotherapy is essential to cure ML-DS whereas most cases of TAM spontaneously resolve without treatment.

To determine the natural history of GATA1 mutations in DS and the true risk of subsequent leukemia conferred by GATA1 mutations we established the Oxford DS Cohort Study that prospectively analyses serial clinical, hematological and GATA1 mutational data from children with DS from birth to age 4y. All children had serial blood counts and smears and mutation analysis of exon 2/3 of GATA1 by next-generation-sequencing (NGS). A diagnosis of TAM was made if peripheral blood (PB) blasts were >10% and a GATA1 mutation was detected. DS neonates with PB blasts <10% and a GATA1 mutation were designated Silent TAM as our previous data on 18/80 DS neonates with blasts <10% showed that small mutant GATA1 clones were clinically and hematologically undetectable [Roberts et al, Blood 2013].

We have now studied 471 neonates (gestational age [GA] at birth 29-42 wks) with karyotypically-confirmed DS. 130/471 (27.6%) had >1 GATA1 mutation confirming the high frequency of GATA1 mutations in DS neonates. Multiple GATA1 clones (2-7) were detected in 19% of TAM and 21% of Silent TAM (p=ns). A diagnosis of TAM was made in 55/471 (11.7%). 75/471 (15.9%) had Silent TAM and 341/471 had no GATA1 mutations. Typical clinical findings of TAM (hepatomegaly, splenomegaly, skin rash and effusions) were uncommon in Silent TAM, affecting 7%, 1.3%, 1.3% and 3% neonates respectively compared to 40%, 29%, 17% and 10% of neonates with TAM (p<0.0001). There were no specific hematological abnormalities in Silent TAM: median WBC and blasts were not increased: 13.1x10⁹/L v 14.5x10⁹/L and 4% v 4% (Silent TAM v no GATA1 mutation) and no neonates with Silent TAM had leucocytosis (WBC >35x10⁹/L) or anemia. There was a strong correlation (r=0.753; p<0.0001) between PB blast% and mutant GATA1 clone size (assessed by VAF) with a greater mutant clone size in TAM v Silent TAM (median VAF 17.1% v 1.2%; p<0.0001).

Unexpectedly, clone size was inversely proportional to GA (r=-0.2569; p=0.004). Despite similar median GA at delivery for all 3 groups, no GATA1 mutations were detected before 31 wks GA and no cases of TAM before 33 wks; the highest frequency of GATA1 mutations was seen at 34-35 wks GA (40%) reducing to 7% at >40 wks. This suggests GATA1 mutations are likely to be acquired and expand during a narrow time window early in the 3rd trimester of fetal life. Consistent with this, GATA1 mutations were not detected in 16 DS fetal liver samples (GA 8-19 wks) analysed over the same period. Similarly, in a case of TAM presenting in utero with fetal pericardial effusion and 96% PB blasts at 27 wks GA, the blast% and mutant GATA1 VAF spontaneously fell progressively (blasts 19%; VAF 23% at birth at 34 wks GA; undetectable at 4 wks postnatal age).

Outcome: 7 neonates developed ML-DS: 6/52 survivors with TAM; 1/71 survivors with Silent TAM and 0/328 survivors without neonatal GATA1 mutations; 2/451 survivors developed ALL (1 Silent TAM, 1 no GATA1 mutation). 5 yr overall survival was the same in all 3 groups (95.1%, 95.8%, 95.1% for TAM, Silent TAM and no GATA1 mutations; P_log-rank 0.8994) while 5 yr event-free survival was lower in TAM (89.2%, 92.8%, 95.1% in TAM, Silent TAM and neonates with no GATA1 mutations; P_log-rank=0.0262). ML-DS was diagnosed at a median age of 14 m (range 3-18 m) preceded in all cases by a fall in platelet count. No single risk factor at birth specifically predicted for ML-DS although all 6 neonates with TAM who developed ML-DS had a mutant GATA1 VAF of >15% and blasts >20% at birth. The DS neonate with Silent TAM had a VAF of 2.5% and 4% blasts; this neonate was the only one of the cohort with a partial trisomy 21. Conclusion: Acquired N-terminal mutations in GATA1 are very frequent in neonates with DS and most probably occur or expand in the 3rd trimester; in the majority of cases mutant GATA1 clones are very small, clinically silent, resolve spontaneously and confer an extremely low risk of ML-DS.