Analysis of Clonal Hierarchy Shows That Other Ancestral Events May Precede Evolution of Del(5q) in Myeloid Neoplasms

The molecular pathogenesis of myeloid neoplasms characterized by 5q deletion (del(5q)) has not been completely elucidated. While some pathomorphologic features including e.g., megakaryocytic and erythroid dysplasia, have been associated with specific genes within minimal common deleted regions (CDR), genes responsible for clonal advantage and expansion have not been identified. It is not clear if haploinsufficiency of one or multiple genes within del(5q) is responsible for clonal evolution or whether mutations in those genes or other genes located in other genomic areas are present. Moreover, with the recognition of intra-tumor diversity and hierarchical clonal architecture, it may be possible to establish whether del(5q) or other lesions, including common somatic mutations, constitute the ancestral event in the pathophysiologic cascade.

We performed a comprehensive mutational screen in 124 patients with del(5q), including 59 patients studied by whole exome sequencing (WES) and 65 by targeted deep NGS of genes within the deleted area and the other most commonly mutated genes as previously determined in WES cohorts. To identify pathogenic genes, those most consistently found to be haploinsufficient in del(5q) were matched for the presence of mutations in diploid cases. For the purpose of this study haploinsufficiency was quantitated based on the number of cases with del(5q) showing <60% expression of the corresponding genes. E.g.,HDAC3 in 81%, PPP2CA in 62% and RPS14 in 14% of cases with del(5q). For all somatic mutations, we also describe the clonal composition based on deep sequencing in serial samples and analyses of variant allelic frequency. Finally, we compare the clonal size for individual mutations with that of del(5q). The latter was accomplished by calculation of clonal size based on allelic imbalance for informative SNPs present within deleted regions in heterozygous configurations in germ line samples. The average deviation from the ideal 50/50 distribution in tumor samples allowed for precise calculation of the proportion of cells in the sample affected by the deletion. Using this approach, there was a good correlation to the size of del(5q) clone by FISH (r=.94)

Our results demonstrate that 10/14 genes were haploinsufficient within the CDR, but only 2 hemizygous somatic mutations were identified. However, 12 mutations in 7 genes (MATR3, SH3TC2, CSNK1A1, PDGFRB, CD74, FAT2 and G3BP1) were present with the area corresponding to the CDR in diploid cases. TP53 mutations were more commonly associated with del(5q) (73%, vs. 27% in diploid 5, p<.001) and were particularly frequent in patients affected with 2 commonly retained regions (CRR1;5q11.1-5q14.2 and CRR2; 5q34-qter), where they were found in 81% of cases (30/37) vs. 19% (7/30) among CDR deletions (p<.001). In lower-risk MDS, mutations were detected in 11% of deletion cases, whereas they were only found in 5% of diploid chr5 (p<.0001). In higher-risk MDS, TP53 mutation were found in 42% of del(5q) vs. 4% of diploid chr5 (p<.0001). Similarly, 45% patients with concomitant -7/del(7q) and del(5q) had TP53 mutations. The most common mutation associated with del(5q) was TP53, while mutations of FLT3, NRAS or TET2 were significantly mutually exclusive (p=0.03, 0.04 and 0.03; respectively). Next we determined the earliest somatic event by comparing of clonal size of the associated lesions. Del(5q) was present in 17-98% of tumor cells. We identified three theoretical possibilities as to the clonal architecture of del(5q) myeloid neoplasms: i) Tumors in which driver somatic mutations precede del(5q) (35%), ii) those in which del(5q) appears to precede any other somatic mutation (6%) and iii) the succession cannot be determined because of very expanded clones of similar size (“clonal saturation”) i.e., these cases were not informative. For cases in which del(5q) was a secondary lesion, TP53 was the ancestral event 64% of the time, and DNMT3A 27% of the time. The TP53 mutation was detected as a secondary event in 1 of 2 samples in which del(5q) was found to be ancestral.

In sum, our results suggest that del(5q) is not universally an ancestral event. The TP53 mutation is the most common mutation in del(5q) and may also serve as ancestral event. While UPD17p and hemizygocity for TP53 can be found in 33% of TP53 mutant cases, most of the detected TP53 mutations were likely to heterozygous, and therefore the clonal size was not overestimated.