Double Genetic Hits and Subclonal Mosaicism in the Ras Signaling Pathway in Myeloid Neoplasia

The high prevalence of oncogenic mutations in the RAS family genes including their down or upstream partners (e.g., NRAS, KRAS, PTPN11, RIT1α) has strong implication in abnormal Ras signaling in the pathogenesis of myeloid neoplasia (MN). For many years RAS mutations have been considered non-targetable. However, recently multiple targeted agents have been introduced as possible future therapies. While common in JMML, RAS mutations are rarely seen in adult MN as the ancestral event. Instead they serve as subclonal hits. To that end, the clinical impact of RAS mutations has been controversial: whereas some studies did not find prognostic impact, others demonstrated that these mutations may confer a poor prognosis. Nevertheless, molecular analysis of core binding factor AML patients showed the coexistence of RAS clones harboring independent hits targeting the same pathway. We stipulate that occurrence of RAS mutations is not random and reasoned that studying cases with multiple RAS mutant subclones may reveal distinct molecular or etio-pathologic background attracting the emergence of RAS mutations as a way to augment MYC activity as a ubiquitous pathway of clonal progression.

In a review of a molecular collection of data of a cohort of MN patients (n=1876) including MDS (n=692), MDS/MPN (n=282), AML (pAML, 710; sAML, 192) we found RAS mutations in 21% (403/1876) of the patients. Among them we encountered a subgroup of patients (9%; 38/403) harboring multiple RAS mutations. These multiple hits affected PTPN11, NRAS, KRAS and RIT1-α. Most of these multiple RAS mutants represented a subclonal mosaicism rather than biallelic subclones as indicated by single cell DNA sequencing. Three patients were assessed serially (2 patients/ 2 time points and 1 patient/ 5 time points). In one case NRAS (VAF-24%) and PTPN11 (VAF-28%) were acquired at the time of AML evolution. In an MDS/MPN-U, NRAS/KRAS were both detected at diagnosis with a VAF of 1% and 12% each. Both clones increased over 2 years' time span with NRAS ramping to 43% and reaching KRAS (50%). In the 3^rd case, samples were collected over a period of 5 years with NRAS and PTPN11 being detected only at the later time point (VAF of 5% and 15%, respectively). In contrast to competing RAS subclones in adults, our historical cohort of JMML showed that at least one RAS mutation was found in 89% (82/92) of the patients as ancestral events and in mutually exclusive manner while biallelic RAS mutations occurred in 15% (12/82) of the cohort.

In adult MN, the most common subclonal mosaicism was encountered in NRAS/KRAS (63%; 24/38), while less fraction of patients had competing subclonal mutations in NRAS (26%; 10/38) or in KRAS (8%; 3/38). Most of these mutations were found in canonical sites (NRAS: G12D/S/A/C, G13D/R/V, Q61A/H/R; KRAS: G12D/A/R, G13D/R/V, A146P/T). Patients carrying RAS mutations had significantly higher WBCs (20.5 vs 7.2, P<.001) consistent with the role of RAS in proliferation, possibly via augmentation of MYC. Isolated chromosome 7 abnormalities were more common among RAS mutant carriers who otherwise showed less complex karyotypes. RAS mutations were mostly enriched in MDS/MPN (26%) while absent in sAML suggesting the impact of these lesions on OS but not on PFS. Analysis of clonal architecture showed that accumulation of RAS subclones was most commonly related to the presence TET2 (17% vs. 8%) and EZH2 (6% vs. 1%) compared to WT. At the time of AML evolution, the presence of RAS mutation did not correlate with OS (18 vs. 13 mo., P=0.07). However, median OS was shorter in MDS and MDS/MPN carrying multiple RAS mutations compared to WT (10 vs. 30 mo., P=0.005) again supporting the contention that these clones may be markers of non-progression-related mortality.

In sum, our study supports the notion that ancestral RAS hits are common in childhood MN with a hereditary component whereby the progression is related to the acquisition of secondary hits in biallelic configuration. In contrast in adult and elderly MN, specific genetic background predisposes to RAS mutant mosaicism such as ancestral TET2 and EZH2 lesions and MPN features reflecting the direction of selection pressure towards accumulation of multiple RAS pathway hits. Lack of the gene/dose effect on progression indicates, that in contrast to the compound heterozygous RAS childhood diseases, RAS mutant mosaicism reflects branching rather than linear evolution mode.