The Clonal Hematopoietic Spectrum of Down Syndrome and ML-DS

Introduction: Children with Down syndrome (DS) have a 150-fold higher risk of developing myeloid leukemia (ML-DS) (Wechsler J, Nat Gen 2002). Ten percent (10%) of DS children are also predisposed to developing a preleukemic self-limiting condition, transient myeloproliferative disorder (TMD). While the majority of these children achieve complete remission, 20-30% of TMD patients will eventually develop ML-DS (Gamis AS, Blood 2011). Unique to ML-DS and TMD is the association with mutations in the X-linked hematopoietic transcription factor, GATA1 (Alford KA, Blood 2011), which are pathognomonic, but perhaps not sufficient, for ML-DS. Along with GATA1 mutations, ML-DS has a distinct mutational landscape implicating the cohesin complex, epigenetic regulators, signal transducers and the RAS pathway (Yoshida K, Nat Gen 2013). While these findings have informed our understanding of ML-DS etiology, ~30% of healthy DS neonates - without overt TMD or ML-DS - also harbor GATA1 mutations (Roberts I, Blood 2013), suggesting that ML-DS arises from clonal outgrowth of these progenitors that acquire specific additional mutations. Because the spectrum of physiologic clonal hematopoiesis is unknown in DS, we sought to characterize clonal mutations in healthy DS children as compared to a cohort of ML-DS cases using our highly sensitive and specific error-corrected sequencing (ECS) methodology. A precise understanding of their altered hematopoietic development may inform risk stratification, therapeutic selection and outcomes.

Methods: The Druley lab has developed a custom error-corrected sequencing (ECS) panel targeting 80 genes frequently mutated in both pediatric and adult myeloid malignancies (Young AL, Nat Comm 2016) that is validated to identify clonal mutations as rare as 0.0001 variant allele fraction (VAF), 100X below the error-rate of next-generation sequencing (Wong TN, Nature 2015; Young AL, Leukemia 2015). Using ECS, we surveyed 102 DS children (47 ML-DS enrolled from a current Phase III ML-DS study; 55 healthy DS enrolled at St. Louis Children's Hospital).

Results: We identified 294 total clonal variants in 72/102 children (60.0%, healthy DS; 83.0%, ML-DS cases) at 0.0002-0.82 VAF. On average, we found 1.8 clonal variants per healthy DS control and 4.1 clonal variants per ML-DS case. Consistent with Yoshida et al., we find a spectrum of recurrent mutations in ML-DS cases: GATA1 (53.2%), EZH2 (25.5%), RAD21 (14.9%), and STAG2 (10.6%) etc. Surprisingly, we also identified frequent mutations in FAT1 (14.9%) and SETD2 (10.6%), genes previously unassociated with ML-DS. Moreover, we report a different set of recurrent mutations in healthy DS children: FAT1 (12.7%), BCOR (12.7%), TET2 (10.9%), SETD2 (7.3%), and TRIM24 (5.5%), and TP53 (5.5%) etc.

Discussion: Earlier ML-DS studies could only detect more common mutations >0.02 VAF due to the error-rate of standard next-generation sequencing (NGS). Using ECS, we present the first characterization of the unique clonal hematopoietic spectrum in children with DS and ML-DS. These results reveal that clonal hematopoiesis is common amongst healthy DS children without hematologic conditions, and that children with ML-DS have a higher clonal mutation burden than those without. Notably, we identified two novel genes recurrently mutated in both ML-DS cases and healthy DS controls: FAT1 and SETD2. We suspect that these mutations were not identified in other ML-DS studies as the median VAFs of our detected FAT1 and SETD2 mutations were 0.001 and 0.0008, respectively. Mutations in FAT1 and SETD2 have been shown to lead to dysregulated Wnt signaling in numerous cancers (Morris LG, Nat Gen 2013; Yuan H, J Clin Invest 2017). We and others have previously shown that Wnt signaling is crucial in specifying a primitive or definitive hematopoietic program (Sturgeon CM, Nat Biotech 2014; Creamer JP, Blood 2017). We speculate that these mutant FAT1 and SETD2 clones, like mutant GATA1 clones, may inhibit the definitive hematopoietic potential in DS children. Lastly, we note a reduced incidence of GATA1 mutations in our healthy DS cohort and attribute this to an older average age (8.3 years), as compared to Roberts et al. studying neonates. Other groups have speculated that somatic mutations in GATA1 (with a DS background) may only occur during a restricted developmental window (<4 years) as TMD clones cycle into quiescence afterwards (Zhe L, Nat Gen 2005; Hasle H, Leukemia 2007).