Molecular Dissection of Del(5q): Distinction between Primary and Secondary Del(5q) and Pathogenetic Implications

Del(5q) is the most common cytogenetic (CG) alteration in myeloid neoplasia (MN), and solo defines the 5q- syndrome. Despite significant progress in understanding the disease mechanisms resulting from del(5q), the pathogenesis behind this lesion, in particular its effect on leukemogenesis, is still not clear. The absence of corresponding somatic LOH in the del(5q) region suggests that the haploinsufficiency (HI) resulting from the deletion might be the key pathogenic factor. However, none of the investigated HI genes located on del(5q) can explain the 5q related growth advantage/transformation. High molecular diversity including somatic mutations in the retained allele, additional CG abnormalities, epigenetic factors (loss/gain of silencing), and LOH for germ line (GL) protective alleles may preclude the identification of the pathogenically essential lesions. We revisit the genomics of del(5q) by taking advantage of a large cohort of patients with molecular (WES, WGS, RNAseq) and clinical annotation. We analyzed a total of 400 samples (388 patients) with del(5q) and 825 diploid patients with MN. Patients were subgrouped into isolated del(5q) (iso-del5q; 49%) and del(5q) with additional lesions referred to as compound del(5q) (comp-del5q).

Availability of paired GL WES results allowed us to reconstruct clonal hierarchy using precise bioanalytic allelic exclusion methods. Studies have postulated that del(5q) is a founder event; we now demonstrate that 52% of iso-del5q patients did have a dominant del(5q) event, but in 56% of comp-del5q and 28% of iso-del5q, del(5q) was subclonal. Co-dominant del(5q) with somatic mutations were also found (iso-del5q: 20% vs. comp-del5q: 11%). When del(5q) was dominant, patients had fewer associated mutations, while cases with secondary del(5q) had a poor prognosis due to ancestral TP53 mutations (MT) and accumulation of chromosomal breaks.

We then focused on the expression level: 405 genes on 5q were interrogated, 188 within q14q34 (CDR-1: 41; CDR-2: 55). We defined HI as expression <25^th %tile of the diploid expression. CSNK1A1 was deleted in 90% and HI in 77% of del(5q), whereas RPS14 was deleted in 89% but HI in 39% of del(5q) cases. Applying a more stringent definition (<68% of diploid), the most consistently HI genes were SIL1 in 61%, H2AFY in 58%, and CTNNA1 in 49% of cases vs. <5% of diploid cases. By this criterion, RPS14, ACSL6 and TGFBI were also HI in 19%, 17% and 16% of diploid cases. Del(5q) HI genes showed an enrichment of 49x in β-catenin phosphorylation cascade genes (P<.0001), 25x in Wnt signaling (P=.001) and 5x in cell cycle genes (P<.0001). When the entire profile was examined, TP53 and apoptotic genes also showed enrichment in upregulation (both P<.0001).

We also genotyped del(5q) and a control diploid cohort for somatic mutations. Within the 5q CDR APC, RAD50, and CSNK1A1 were most frequently mutated (all hemizygous), particularlly in iso-del5q. No canonical DDX41 frame shift mutations (GL) or somatic mutations occurred in its ATP binding domains were found. However, there were DDX41^MT (n=2) that coincided with del(5q) and one patient coincided with the CDR. Del(5q) patients had a distinct mutational profile of co-associated lesions with a higher frequency of TP53^MT (P<.0001) and a significantly lower frequency of SF3B1, ASXL1, TET2, JAK2, SETBP1, U2AF1 and SRSF2 mutations (all P<.01) than diploid patients. TP53^MT were enriched in comp-del5q in cases that frequently also demonstrated -7/7q- and 17p-.

CSNK1A1 and TP53 are the two most common mutations in del(5q) patients. CSNK1A1^MT were co-mutated with TP53 in only 15% of patients and CSNK1A1 HI did not enrich for TP53^MT. Expression of p21 (a TP53 activation marker) was up-regulated in del(5q) except for cases with biallelic TP53 inactivation. RPS14 was HI in a fraction of del(5q) patients. Patients with the RPS14 deleted locus were more likely that those with HI to have an increase in p21 expression (P<.0001) but only in patients with advanced disease.

In sum, our analysis of a comprehensive compendium of del(5q) genomics revises previous assumptions (del(5q) is not a uniform founder lesion, and the heterogeneity HI of genes across patients), discovers new HI candidate genes and precisely describes molecular relationships (e.g., del(5q) with TP53) and generated a minimalistic expression signature of del(5q).