Whole Exome Analysis Of Germline Alterations Associated With Myelodysplastic Syndrome

While in the past germline (GL) events have been investigated in myeloid disorders, these studies focused on targeted genotyping of empirically selected candidate genes, and a comprehensive characterization of GL-encoded susceptibility has not been undertaken in these diseases. Whole exome sequencing (WES) data sets obtained in the molecular discovery projects, allow for a targeted or unbiased analysis of inherited alterations to identify those with a pathogenic importance. To that end, initial WES studies, including published analyses from TCGA leukemia cohort, have focused mainly on somatic mutations.

MDS is a disease of the elderly, typically presenting with complex phenotypes and clinical histories. Inherited predisposition, if present, would be expected to follow complex, non-mendelian genetic traits, have low penetrance and interact with various extrinsic factors such as exposures or coexisting conditions, and family histories may not be informative. Therefore, in this disorder predisposing GL alterations may be difficult to discern.

Here, we have applied WES in patients with MDS and other myeloid neoplasms (N=117). We also investigated AML data set available through TCGA (N=201). In addition to defining somatic mutations in paired (tumor, GL DNA), we also performed an unbiased search for GL SNPs and mutations.

Our strategy was comprehensive; firstly it focused on selection of non-synonymous and, possibly deleterious SNPs. Subsequently, putative candidates were further prioritized based on their genotypic frequency in the general population, and finally, according to perceived importance for the leukemogenesis. For the purpose of this study, we used a maximum minor allele frequency threshold of .05 in matched general population for alterations to be prioritized.

As a result, we identified a large variety of new and, as expected, previously reported GL polymorphisms and mutations. Known pathogenic as well as novel GL alterations (42 non-synonymous sites, reported (N=33) and novel (N=9)) were found in telomerase genes (TERT, DKC1, GAR1, POT1, SMG6, NOP10, TINF2, NHP2, WRAP53) and other bone marrow failure genes (ELANE, GFI1, HAX1, GATA2, CSF3R, WAS) or many others.

Subsequently, we hypothesized that there may also be GL variants of pathogenic significance in genes that are also frequently affected by somatic events. Known examples of such genes include SETBP1, NF1, CBL and many others. Such GL events may be directly disease-prone or indirectly predispose to somatic lesions within the same gene (as recently shown for JAK2 V617F mutation). Since the complexity of the entire analysis exceeds the boundaries of this abstract we focus here on exemplary results of genes most frequently affected by somatic mutations, including e.g., TET2 (8%), ASXL1 (15%), DNMT3A (26%), CEBPA (6%), TP53 (8%) and others. In this illustrative subset alone, 647 non-synonymous variants across 74 sites, in 14/16 genes of interest were found (FLT3, TET2 and TP53 were most frequently affected). Following, bio-analytic filtering we identified 30 rare variants, of which 11/30 were present at a significantly higher frequency compared to control population. Of these, 5/11 were located within such gene as ASXL1: S737N (p=.002), P1213R (p=.005), G543S (p=.023), L1286V (p=.026), L1216F (p=.036), 3/11 in FLT3: D324N (p<.001), A912V (p=.002) ,F298L (p=.003), the remaining were in DNMT3A: R693H (p=.042), KIT: N293S (p=.019), and KRAS: C180X (p=.002). Of note is the FLT3 polymorphism, D324N, which has been previously linked to AML susceptibility. Excluding reported polymorphic sites, 33% of non-synonymous sites were not reported in population databases.

Our data identified several GL variants that appear to associate with specific phenotypic features. For instance, a significantly decreased overall survival was seen in patients who had a RUNX1/RUNX1-IT (p=.03) GL mutation. Furthermore, patients diagnosed with MDS/MPN were found to have a greater odds of presenting with concurrent GL and somatic variants in TET2 gene compared to non-MDS/MPN patients [OR 7.4; 95% CI 2.5-21.8).

In conclusion, our data suggests that both GL and somatic events in myeloid disorders are important for the pathogenesis of myeloid malignancies. They may interact with each other and with cytogenetic abnormalities, which can lead to deletion of protective alleles or amplification of disease-prone alleles.