Novel and Significant Impact of Germline Variants Predisposed to Pathogenic Somatic Mutations and Loss of Heterozygosity (LOH) in Myelodysplastic Syndromes (MDS) and Clonal Hematopoiesis of Indeterminate Potential (CHIP)

While germline predisposition to myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML) has long been recognized mainly through rare familial and pediatric cases, it has been drawing an increasing attention, on the basis of the recent discovery of novel risk alleles for MDS/AML through studies relying on revolutionized sequencing technologies; according to these studies, it suggest that more numbers of MDS/AML cases than expected might have germline predisposition. Moreover, it is suggested that germline variations may also confer predisposition to age-related clonal hematopoiesis or "CHIP", which has been implicated in the development of MDS/AML.

In this study, we explored germline predisposition to MDS and CHIP through intensive sequencing of blood samples from large cohorts of AML/MDS patients and 'hematologically' healthy individuals (HHIs), in which germline variants in 21 genes implicated in sporadic or familial MDS/AML or CHIP were interrogated among patients with MDS/AML from the Japan Marrow Donor Program (n=797) and HHIs aged >60 years from Biobank Japan (n=10,852). Germline variants were referred to NCBI dbSNP Build 151 database, excluding the entries in COSMIC ver.7 and in-house database, followed by manual curations. Somatic mutations and CHIP in the 21 genes were also analyzed for MDS/AML and HHIs, respectively.

In total, 30,286 germline variants, including both synonymous and non-synonymous changes, were detected in 21 genes in the entire cohort. By comparing their frequencies between in MDS/AML and HHIs, we identified 6 germline variants in showing a significant enrichment in MDS/AML. Among these most frequently observed was variants in DDX41, for which a total of 3,721 variants were detected in 3,688 HHIs. Among these, 3 variants were significantly enriched in MDS/AML, including p.A500fs (OR=13.1 [6.6-25.9] (95%CI) (n=15), p.S363del (OR=41.0, [4.3-349.5]) (n=3), and p.Y259C (OR=34.2, [6.6-176.8]) (n=5). Of interest, 14 of 23 MDS patients with one of these alleles carried somatic DDX41 mutations, typically p.R525H, which were not found in any of HHIs, further supporting the relevance of these DDX41 risk alleles. Also including an additional 2 nonsense/splicing variants, 5 DDX41 alleles found in 25 MDS/AML patients were thought to represent germline predisposition to MDS/AML. Similarly, RUNX1 p.H85N (OR=9.10, [1.52-54.52]) (n=2), CBL p.P782L (OR=4.27, [1.56-11.70]) (n=5), and GNAS p.H69N (OR = 2.90, [1.28-6.59]) (n=7) showed a significant enrichment in MDS/AML. Combined, these putative risk alleles accounted for 4.6% (37/797) of sporadic MDS and sAML. None of these alleles were observed in the Caucasian population of Exome Aggregation Consortium dataset, suggesting Asian origins of these variants.

We next evaluated the effects of germline variants on CHIP. CHIP mutations were detected in 929 HHIs, where DNMT3A mutations (n=290) were most prevalent, followed by TET2 (n=124) and ASXL1 (n=68) mutations. By comparing allele frequency of each of 1,276 germline variants between healthy donors with and without CHIP, we identified two haplotypes at the JAK2 and TET2 loci, defined by T/A at c.C489T/c.G2490A (JAK2) and G/G/T at c.G652A/c.G3117A/c.T4140C (TET2), which were significantly enriched in the cases carrying CHIP with the JAK2 (p.V617F) and TET2 mutations, respectively (T/A vs. C/G; OR=3.36, [1.41-8.01] for JAK2 and G/G/T vs. A/A/C; OR=1.85, [1.19-2.86] for TET2). Intriguingly, the JAK2 risk haplotype (C/G) were also enriched in MDS cases with JAK2 p.V617F mutations (T/A vs. C/G; OR=3.06, [1.26-7.60]). Similarly, the TET2 risk haplotype (G/G/T) tended to be enriched in MDS cases with TET2 mutations, although not statistically significant. Finally, variant allele frequency of JAK2 p.V617F mutations in CHIP exceeded 0.5 in 4 out of 26 JAK2 CHIP-positive patients (15%), suggesting the presence of loss of heterozygosity (LOH) in chromosome 9p.

In conclusion, through a large-scale detection of germline variants in 21 common drivers of MDS/AML as well as CHIP, we identified multiple novel germline variants or haplotypes that showed a significant predisposition to the development of adult-onset MDS or CHIP, respectively. Our findings provide novel insights into the genetic basis of myeloid leukemogenesis and the development of CHIP.