The complex genetic landscape of familial MDS and AML reveals pathogenic germline variants

Abstract

The inclusion of familial myeloid malignancies as a separate disease entity in the revised WHO classification has renewed efforts to improve the recognition and management of this group of at risk individuals. Here we report a cohort of 86 acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS) families with 49 harboring germline variants in 16 previously defined loci (57%). Whole exome sequencing in a further 37 uncharacterized families (43%) allowed us to rationalize 65 new candidate loci, including genes mutated in rare hematological syndromes (ADA, GP6, IL17RA, PRF1 and SEC23B), reported in prior MDS/AML or inherited bone marrow failure series (DNAH9, NAPRT1 and SH2B3) or variants at novel loci (DHX34) that appear specific to inherited forms of myeloid malignancies. Altogether, our series of MDS/AML families offer novel insights into the etiology of myeloid malignancies and provide a framework to prioritize variants for inclusion into routine diagnostics and patient management.

Fig. 1. Group 1 families: Families with variants in established genes.

a Number of families with variants in known disease-causing loci. In gray, genes where the level of evidence for gene–disease association is high; moderate level of evidence in blue; and genes emerging from basic research or mutated in other inherited hematological syndromes with high risk of MDS/AML in red. Variants in these genes are heterozygous, apart from ERCC6L2, FANCA, and SBDS where they are biallelic, and WAS where it is hemizygous. b Percentages of our family cohort with causal variants in known disease-causing genes.

Fig. 2. Unreported families with variants in established loci.

Pedigree representation of the 17 newly described families that include one additional CEBPA family (c.1138_988dupGAAC:p.Gln330Argfs*74) (a), two DDX41 (c.3G>A:p.Met1? and c.370C>T:p.Arg124X) (b, c), one ETV6 (c.349C>T:p.Leu117Phe) (d), two GATA2 (c.1061C>T:p.Thr354Met and c.1084C>T:p.Arg362X) (e, f), three RUNX1 (RUNX1 deletions: del(21):36349450–36572837, del(21):36400658-36972948, and del(21):36389492-37056053) (g–i), one SAMD9L (c.4418G>A:p.Ser1473Asn) (j), one TERT (c.1445delA:p.His482Profs*27) (k), one TERC (r.179_180TCdelinsGG) (l), one MECOM (c.2443C>T:p.Arg815Trp) (m), one FANCA (c.2505-1G>T/c.3626+5G>C) (n), one SBDS (c.258+2T>C/c.183_184delinsCT:p.Lys62X) (o), one WAS (c.1336delA:p.Lys446fs) (p), and one TP53 (c.844C>T:p.Arg282Trp) (q). Age of the patients is indicated as a number in gray font color and italic. Index case of each family is indicated with an arrow.

Fig. 3. Summary of disease genes and new candidate genes mutated in our cohort of MDS/AML families.

Schematic representation of established disease genes (Group 1) and new candidate genes (Group 2) that are mutated in our cohort of MDS/AML families, and their overlap with classical inherited BMF syndromes based on Bluteau et al. classification (blue) or other inherited hematological disorders (pink). In orange font, genes reported to be frequently mutated in sporadic AML.

Fig. 4. DHX34 variants fail to phosphorylate UPF1.

a Cartoon depicting the domain structure of DHX34 indicating the position of the reported variants. b HEK293T cells depleted of DHX34 with a specific siRNA or transfected with a scrambled non-targeting siRNA (−) were co-transfected with FLAG-UPF1 and siRNA-resistant wild-type (WT) T7-DHX34 or the indicated DHX34 variants (including an empty vector plasmid (−) as a control on lanes 1–2). The catalytically inactive DHX34 mutant K191A served as a negative control (lane 9). Phosphorylated UPF1 was detected with a phospho-(Ser/Thr) ATM/ATR substrate antibody and the phospho-FLAG-UPF1 signal was normalized to the levels of UPF1 recovered in the IP. Inputs (0.5%) (upper panel) and Anti-FLAG-immunoprecipitates (20%) (lower panel) were probed for the indicated proteins. c Quantification of the western blot signal, as shown in panel b. Mean values ± standard deviations from at least two independent experiments are shown. d Analysis of T7-DHX34 (wild-type protein (WT) and variants) binding to FLAG-UPF1. HEK293T cells were transfected with wild-type T7-DHX34 or DHX34 variants and FLAG-UPF1. Inputs (0.5%) and Anti-T7-immunoprecipitates (20%) were probed for the presence of UPF1. Uncropped western blots are provided as a Source Data file.