A prospective Study of whole genome sequencing for genomic profiling and risk determination in MDS

Abstract Introduction: Genetic risk stratification in MDS is critical to predict disease outcomes and to direct therapy. Our group has developed a whole genome sequencing (WGS) approach for comprehensive genomic analysis at presentation in myeloid malignancies. We hypothesized that WGS as a single platform would be feasible in a real-time clinical setting and would provide increased diagnostic and prognostic yield compared to conventional genomic profiling using cytogenetics (metaphase karyotyping without FISH) combined with targeted gene panel sequencing (i.e., targeted sequencing+cytogenetics). Here, we report a prospective clinical trial to determine feasibility of WGS incorporation into real-time clinical evaluation of MDS. Methods: WGS was performed on DNA isolated from bulk peripheral blood or bone marrow aspirate samples to a target depth of 60x without paired normal DNA sequencing. Targeted analysis was performed for small variants in 64 recurrently mutated myeloid genes and genome-wide identification of copy-number alterations (CNAs) > 5 Mbp, copy neutral loss of heterozygosity (cnLOH), and structural variants (SVs) involving 639 genes. Conventional genomic profiling consisted of high-coverage (>1000x) targeted gene panel sequencing of the same 64 genes with a sensitivity of 2% variant allele fraction (VAF), that includes all IPSS-M genes. Metaphase cytogenetics was performed by institutional standards. Results: We prospectively enrolled 66 patients. Conventional cytogenetics was unsuccessful or inconclusive in 9 (14%) patients. In contrast, WGS was successful in 100% cases and 17 of 19 (89%) cases categorized as complex by cytogenetics were also identified as complex by WGS. In cases with noncomplex cytogenetic changes, WGS identified 100% of SVs (2 of 2) and 89% of CNAs (19 of 21) identified by cytogenetics. WGS also identified new cnLOH and CNAs in 11 of the 57 cases with sufficient cytogenetics, including cnLOH of chr 17p (n=2) and del 17p (n=1), which led to new identification of multihit TP53 in these 3 cases. In 9/9 cases with unsuccessful/inconclusive conventional cytogenetics, WGS provided cytogenetic information, permitting IPSS-M risk categorization as follows: Very Low (n=1), Low (n=3), Moderate Low (n=1), High (n=1), Very High (n=3). In 49 of the 57 cases (86%) with sufficient cytogenetics, the IPSS-M risk categories were concordant between WGS and conventional genomic profiling. In 1 case, WGS stratified a patient into higher risk categories due to new identification of a complex cytogenetic risk category. In 7 cases, WGS stratified patients into the next lower IPSS-M risk category compared to targeted sequencing and cytogenetics due to limitations of detecting low abundance IPSS-M relevant mutations using lower coverage WGS (12 variants with a median VAF of 4.3% [range 3.7-10.8][n=6 cases]) or low abundance or missed CNAs ([<10% cell fraction][n=3 cases, 2 also with missed variants]). However, the VAFs of all mutations detected by both WGS and high-coverage targeted sequencing were significantly correlated (r=0.91, n=185 variants, p<0.0001). To overcome the limitations of lower coverage WGS to detect sequencing variants, we next combined variants detected by either WGS or high-coverage targeted sequencing (WGS+targeted sequencing) and compared IPSS-M risk categories in 57 cases with sufficient cytogenetics. WGS+targeted sequencing stratified 1 patient into a higher risk and 1 patient into a lower IPSS-M category (due to missing low abundance CNAs) compared to targeted sequencing+cytogenetics. This suggests that deeper WGS coverage may enhance variant detection and further improve risk stratification when using WGS as a single assay for IPSS-M risk categorization. Conclusion: WGS is feasible in MDS and has several advantages over conventional genomic profiling, including when cytogenetics fail or are inadequate, and detection of cytogenetically silent CNAs and cnLOH. Compared to conventional genomic profiling, WGS provided new genomic information that changed IPSS-M risk categories in 10 (15%) of patients. The current cost of reagents, labor, and analysis for WGS described here is comparable to conventional testing. Increasing WGS coverage to >90x will improve detection of mutations with VAFs<10%, making WGS an ideal single assay for IPSS-M risk stratification. Further studies with outcomes data are needed.