Clonal Mutational Landscape of Childhood Myelodysplastic Syndromes

Childhood myelodysplastic syndromes (MDS) belong to a rare group of disorders of aberrant clonal hematopoiesis manifesting throughout entire childhood and adolescence. We had previously established that GATA2 germline mutations can be considered the most common "first hit" in pediatric MDS seen in 7% of primary MDS. However the secondary somatic aberrations facilitating leukemogenesis are not elucidated in children. Previous sequencing efforts established that most somatic mutations very frequently encountered in adults, i.e. affecting TET2, DNMT3a, and the spliceosome genes, do not play a role in the pathogenesis of childhood MDS. Here we aim to define the global mutational landscape in childhood MDS using targeted next-generation sequencing (NGS) approaches.

We investigated children and adolescents enrolled in the prospective studies of the European Working Group of Childhood MDS. Diverse target enrichment and NGS strategies were established including hybridization capture and Ampliseq PCR, Illumina Miseq/Hiseq and Iontorrent PGM. We first examined a pilot cohort of 68 patients for mutations in 138 myeloid leukemia genes. This allowed for the identification of recurrently mutated genes that were selected to be included in a pediatric MDS panel encompassing 28 genes. Targeted NGS using the Iontorrent PGM identified known recurrent mutations. However, the high indel error rate and coverage gaps in homopolymeric regions i.e. in ASXL1 precluded further studies. Using inhouse-adapted Ampliseq-Miseq approach we then sequenced DNA from bone marrow of 586 MDS patients (469 primary and 117 secondary MDS after radio/chemotherapy or inherited bone marrow failure syndromes) at an average depth exceeding 700 reads per amplicon. Somatic mutations were identified in 22% of primary MDS patients, with 1, 2 and 3 genes affected in 16%, 4.5%, and 1.5% of cases, respectively. In secondary MDS twice as many patients (46%) carried mutations; 1, 2, and 3 genes were concurrently mutated in 32.5%, 9.5%, and 4% of patients, respectively. Longitudinal NGS analyses and single CFU colony sequencing confirmed the presence of multiple somatic clones evolving in a hierarchical manner throughout disease course. Most frequent mutations identified in more than 1% of our study cohort of primary MDS were: SETBP1 (7%), ASXL1 (6%), NRAS/KRAS (5%), RUNX1 (3%), PTPN11 (3%) and BCOR/BCORL (1.5%); and in secondary MDS: RUNX1 (14.5%), TP53 (9%), NRAS/KRAS (8.5%), ASXL1 (8%), SETBP1 (6%), PTPN11 (6%), CBL (5%), BCOR/BCORL1 (3.4%). Other genes mutated at very low frequency of <1% were CTCF, STAG2, RAD21, PTEN, JAK3, FLT3, CSF3R, and EZH2. No somatic mutations were identified in CEBPA, GATA1, GATA2, JAK2, cKIT, VPS45, MPL, CALR, SH2B3. The analysis of individual mutational clusters established that for some mutations strongly associate with monosomy 7; i.e. this cytogenetic subgroup was prevalent in 100% of EZH2, 90% of SETBP1, 79% of RUNX1, and 74% of ASXL1 - mutated children with MDS

In conclusion, we established and cross-platform validated a targeted NGS-panel for pediatric MDS, allowing to identify clonal mutations at a sensitivity of at least 10%. In this, to our knowledge largest systematicaly studied cohort of children with MDS we show that mutational load and clonal complexity differs between primary and secondary MDS, and maintains specific patterns for monosomy 7. Although the biological significance of these genomic changes is currently not understood, the emerging distinctive patterns may already be helpful in establishing therapeutic subgroups.