Clinical “MUTATOME” Of Myelodysplastic Syndrome; Comparison To Primary Acute Myelogenous Leukemia

Chromosomal aberrations and somatic mutations constitute key elements of the pathogenesis of myelodysplastic syndromes (MDS), a clonal hematologic malignancy characterized by cytopenias, a dysplastic bone marrow and propensity to clonal evolution. Next generation sequencing (NGS) enables definition of somatic mutational patterns and clonal architecture as a discovery platform, and for clinical applications.

We systematically applied NGS to 707 cases of MDS and MDS-related disorders. 205 cases (low-risk MDS: N=78, high-risk MDS: N=42, MDS/MPN: N=48 and sAML: N=37) were tested by whole exome sequencing (WES). For validation in an additional 502 patients (low-risk MDS: N=192, high-risk MDS: N=104, MDS/MPN: N=111 and sAML: N=95), targeted deep NGS was applied for 60 index genes which were most commonly affected in the cohort analyzed by WES. For NGS data analysis a statistical pipeline was developed to focus on: i) identification of the most relevant somatic mutations, and ii) minimization of false positive results. We studied serial samples from 21 exemplary informative patients. We also compared somatic mutational patterns to those seen in primary AML TCGA cohort (N=201). Given the size of the cohort, there was, for example, a 87% chance of seeing mutations at a frequency of 1% and a 98% of seeing those with a frequency of 2%. While focusing on the most common events, we observed 1117 somatic mutations in 199 genes. The 88 genes mutated mutated in >1% of cases with MDS carried 388 mutations in MDS+sAML (2.5/case), 128 in MDS/MPN (2.7/case) and 398 in pAML (2.0/case). The average number of mutations per case increased during progression (2.2 in lower-risk, 2.8 in higher-risk MDS, 3.4 in sAML). In MDS, the 30 most frequently affected genes were present at least once in 70% of patients. The 30 most frequently mutated genes in MDS/MPN were mutated in 82% of patients. Individual mutations were also sub-grouped according to their function. When we compared three MDS subcategories (lower-risk, higher-risk MDS and sAML) in a cross-sectional view, RTK family, RAS family, IDH family and cohesin family mutations were more frequently detected in the sAML group than in the MDS group. In contrast, the frequency of the DNMT family, TET2 and ASXL family gene mutations did not increase in frequency in the sAML cohort. In addition to better definition of mutational patterns of known genes, we have also defined new mutations, including in the RNA helicase family and the BRCC3pathway.

Clonal architecture analysis indicates that mutations of TET2, DNMT3A, ASXL1, and U2AF1 most likely represent ancestral/originator events, while those of the IDH family, RTK family and cohesin family are typical secondary events. Establishment of mutational patterns may improve the precision of morphologically-based diagnosis. The comparison between MDS-related diseases (MDS+sAML) and pAML revealed a notably different mutational pattern suggestive of a distinct molecular derivation of these two disease groups. While RTK, IDH family and NPM1 mutations were more frequently observed in the pAML cohort, mutations of SF3B1 and SRSF2, were more common in MDS+sAML. With regard to the connections between individual mutation combinations, RTK mutations were strongly associated with DNMT, but not with RAS family mutations in the pAML cohort, while the mutual association between TET2 and PRC2 family, cohesin family and RUNX1were encountered in the MDS+sAML cohort.

Individual mutations may have prognostic significance, including having an impact on survival, either within the entire cohort or within specific subgroups. In the combined MDS cohort, TP53 family mutations were associated with a poor prognosis (HR; 3.65, 95%CI; 1.90-7.01, P<.0001) by univariate analysis. Similar results were found for mutations in TCF4(HR; 7.98, 95%CI; 1.58-10.1, P<.0007). Such an individual approach does not allow for assessment of the impact of less common mutational events.

In conclusion, our study continues to indicate the power of NGS in the molecular analysis of MDS. MDS and related disorders show a great deal of pathogenetic molecular overlap, consistent with their morphologic and clinical pictures, but also distinct molecular differences in mutational patterns. Some of the specific mutations are pathognomonic for specific subtypes while some may convey a prognostic rather than discriminatory value.