Elucidating the Somatic Genetic Rescue Underlying Del(20q) Myeloid Neoplasms

Deletion of the long arm of chromosome 20 (20q-) is a common lesion in myeloid neoplasia (MN). In Shwachman-Diamond syndrome (DBA), 20q- can evolve as a somatic genetic rescue (SGR) event, but in adult patients with 20q-, germline alterations were not identified. When we previously queried alterations in other inherited BMF genes, no significant hits were identified to explain emergence of this lesion in adult MN as a form of SGR (1). Similarly, 20q- did not appear to represent loss of heterozygosity (LOH) from heterozygous germline mutations.

Herein, we aim to examine somatic variants within and outside the 20q- common deleted region (CDR) to narrow the search for culprit genes associated with the pathogenesis and resultant phenotypes of 20q-. This approach will help identify LOH of potential tumor suppressor genes (TSGs) for which deletion would remove the protective allele.

Upon review of 4,751 MN patients, we identified 136 cases with del(20q). 79 of the del(20q) patients had sufficient molecular annotation including SNP array, RNASeq and NGS for identification of somatic hits. SNP array was used to create a cartography of the CDR, defined as bands deleted in >50% of patients and deemed at q11.22-q13.13. From the 445 genes within the CDR, the expression of 392 was restricted to hematopoietic cells. Results from log2 fold change showed that while 100 genes were found to have decreased expression vs diploid controls, the clonality of the 20q- inversely correlated with expression of 64 genes allowing for further refining of putative culprits affected by haploinsufficiency. Notably, EIF6, the hallmark gene of SGR in DBA, was haploinsufficient (HI) but did not correlate with increased clonality. There were no genes within the CDR that paradoxically increased with increased clonality suggesting that LOH of a protective allele of oncogenes is not a possibility. From the 64 genes, 7 were known TSGs: RBL1, E2F1, NCOA5, PTPN1, ZMYND8, STK4, and MYBL2. Notably, MYBL2 is a synthetic lethal gene which, if ablated, in mice led to clonal dominance with a resultant MDS phenotype. There were no suspicious, functionally impactful SNPs in 6 of the 7 identified TSGs. In PTPN1, 11 rare SNPs were identified but no allelic imbalance for a minor allele was found. Unsupervised hierarchical clustering of the 64 HI genes correlating with clonality resulted in 2 distinct clusters. There were no clusters resembling 20q- among diploid cases. Further non-linear dimension reduction by UMAP revealed a clear distinction in MDS vs MPN suggesting diverse gene expression patterns associated with each diagnosis (p=.002).

To identify potential synthetic lethal targets, we then studied HI genes in the CDR from prior studies to examine for lethality in knockout (KO) murine and cell line models: only L3MBTL1 and MYBL2 were identified with the former not correlating with clonality.

Further, we searched for genes on other chromosomes showing consistent compensatory up or down regulation in correlation with 20q. From the 12,583 genes, 172 genes were significantly different between del(20q) vs diploid samples. 143 of the genes were up modulated (17 with known small molecule inhibitors) and 19 were down modulated (none known as synthetic lethal in KO models). We further limited the 172 genes by correlation with clonality: 17 genes had significant correlation (r² > 0.89). 5 of 17 were known to be important to erythroid differentiation or leukemia: LDB1, RHOA, TCF3, SLC20A1, and HEXIM1. HEXIM1 was identified as a previous synthetic lethal gene with a key role in erythroid non-clonal expansion in a murine model.

Hierarchical clustering of non-20q genes with altered expression in 20q- showed 2 clusters: one composed of 3 del(20q) patients with clonality ranging from 40 to 60% and a second with the remaining del(20q) and diploid patients. Further unsupervised subclustering found one cluster (from ten total) with 78% of del(20q), 33% of HR MDS, 27% of LR MDS, and 7% of MPN patients which associates del(20q) more to a diploid MDS than a MPN.

In summary, we found 7 TSGs within the CDR and 5 outside of it that may be culprit genes for the del(20q) MN phenotype. There were 2 synthetic lethal target genes (MYBL2, HEXIM1). Experimental analysis of these two will be performed to assess their role as therapeutic targets. Additional gene enrichment and coexpression network analysis of the non-20q genes and diploid MDS samples subclustered with del(20q) is ongoing.

Maciejewski:Alexion: Consultancy, Honoraria; Novartis: Consultancy, Honoraria, Speakers Bureau.