.AML relapse
We performed unsupervised hierarchical clustering analysis for 46 drug ex vivo sensitivity on 29 primary AML samples from diagnosis and relapse and found a "sensitive" group to several tyrosine kinase inhibitors (TKIs). We focused the rest of our analysis on the multi TKI, dasatinib and searched for correlations between dasatinib response, exome sequencing and gene expression from our dataset and from the Beat AML dataset. Mutations in FLT3-ITD and PTPN11 were enriched in the dasatinib sensitive samples as oppose to mutations in TP53 which were enriched in resistant samples. Gene expression analysis identified that LYN, HCK, CSK and EPHB2 genes were significantly overexpressed in the dasatinib responders carrying FLT3/ITD. Expression levels of these genes could predict which carriers of FLT3/ITD will respond to dasatinib. Based on our prediction, we selected FLT3/ITD AML samples and injected them to mice. Our results demonstrate that in a subgroup of FLT3/ITD AML with a unique gene expression signature, dasatinib significantly inhibited LSCs engraftment1.
2. Predicting progression to AML
To understand which individuals will progress from CH to AML we identified 95 patients with AML from a previously studied prospective European population cohort of >500,000 individuals from whom samples were collected on average 6.3 years before a diagnosis of AML (pre-AML). From the same cohort, we studied 414 well matched unaffected individuals. We found2 that in the pre-AML group, preleukemic mutations and CH were more common than in unaffected individuals. The risk to progress to AML was associated with an increase in both the clone size and median number of mutations per individual. More importantly we identified that mutations is specific genes such as SRSF2 and U2AF1 were associated with high risk to develop. Our manuscript was cover by a news and views manuscript 3.Together with Amos Tanay lab from the WIS, Clalit data4, and single cell RNA sequencing projects, we received preliminary results from the EPIC cohort that pre-AML cases had increased red cell distribution width (RDW) in comparison to controls. Next, we analyzed the electronic health record from Clalit, which contained dynamic clinical attributes of nearly 4.4 million members over 15 years. The analysis confirmed that the increase in RDW preceded AML development by several years. Inspired by this association, we used a machine learning approach to create an AML risk prediction model that relied entirely on routinely available dynamic lab attributes. The model performed with 25.7% sensitivity and 98.2% specificity in predicting AML development 6 to 12 months earlier. As written by one of the editorials about this study “the findings reported by Shlush and colleagues undoubtedly are paradigm shifting in AML, which has long been considered as an unpredictable and unpreventable disease. It will also generate an enthusiasm toward the possibility of AML prevention through early intervention in a high-risk population, although this will require a completely different level of preclinical studies and discussions”5
3. Development of novel deep targeted sequencing technologies
In the study by Biezuner T. et.al 6 we developed new chemistry and novel analytical tools for the calling of low VAF mutations under the improved MIP (iMIP) protocol we developed. For the first time we have validated low VAF mutations called by our iMIP chemistry by a different library preparation protocol and used machine learning methods to call true variants from noise. We also developed the infiniseq technology which allowed us to produce our own panels cost effectively and commercialized this technology to a startup company we founded named Sequentify LTD.
4. Pseudo-mutant P53 is a unique phenotype of DNMT3A-mutated pre-leukemia.
By combining single cell analyses and P53 conformation-specific monoclonal antibodies we studied preL-HSPCs from primary human DNMT3A-mutated AML samples. We found that while leukemic blasts express mainly the WT conformation, in preL-HSPCs the pseudo-mutant conformation is the dominant. HSPCs from non-leukemic samples expressed both conformations to a similar extent. Treatment with a short peptide that can shift the dynamic equilibrium favoring the WT conformation of P53, specifically eliminated preL-HSPCs that had dysfunctional canonical P53 pathway activity as reflected by single cell RNA sequencing. Our observations shed light upon a possible targetable P53 dysfunction in human preL-HSPCs carrying DNMT3A mutations. This opens new avenues for leukemia prevention7.
5. Modelling of ASXL1 deletions by CRISPR/CAS9 exposed novel mutation mechanisms in clonal hematopoiesis.In Feldman T. et.al 8 we discovered that recurrent somatic deletions in human myeloid malignancies (AML, MDS, MPN, CMML) share a similar mutational signature (microhomology (MH)-based deletions). We provided experimental evidence that the recurrent MH-based deletions in ASXL1 SRSF2 and CALR originate in long lived multipotent stem cells. Additional data analyses suggest that the enrichment of the recurrent MH-based deletion in ASXL1 gene might be the result of specific mutational mechanisms in addition to the selective advantage they provide. Our experimental data indicated that the recurrent MH-based deletions in ASXL1 and SRSF2 genes could be successfully recapitulated in vitro in five different cell lines and in primary CD34 cells. By using different knock out models and inhibitory compounds treatments, we demonstrated that these deletions are generated by PARP1 dependent and polymerase theta independent microhomology-mediated end-joining (MMEJ) repair. Combined analyses of single human HSPCs RNA-seq data together with aphidicolin treatment suggested that the sub-pathway leading to preleukemic microhomology mediated end joining (preL-MMEJ) deletions is active mainly during cell replication and might be mediated by replication associated polymerases, a model previously relatively ignored.
1. Tavor S, et al. Haematologica. 2020;105(12)
2. Abelson S, et al. Nature. 2018;559(7714):400–404.
3. Sellar RS, Jaiswal S, Ebert BL. Nature Medicine. 2018;24(7):904–906.
4. Cohen NM, et al.. Nature Medicine. 2021
5. Takahashi K. Cell Stem Cell. 2018;23(2):162–163.
6. Biezuner T, et al. NAR Genomics and Bioinformatics. 2022;4
7. Tuval A, et al. Haematologica. 2022
8. Feldman T, et al. Nature Communications. 2021;12(1)
