Germline alterations of hematopoietic growth factor receptors and maladaptive somatic genetic rescue leads to myeloid neoplasia Free

Using rationally targeted NGS screen, we have previously identified germ line (GL) alteration in genes linked to hereditary bone marrow disease in 5% of myeloid neoplasia (MN) and 10% of bone marrow failure (BMF).¹ In BMF, somatic gene rescue (SGR) may explain leukemic predisposition linking genetic pathogenesis of BMF and MN. Prominent examples include EIF6 or TP53 mutations in SBDS among others. In this context, somatic CSF3R gain of function (GOF) mutations in severe congenital neutropenia (SCN) due to ELANE mutations provide illustration of SGR through hypermorphic mutations (e.g., growth factor receptors; HGFR). We here hypothesize that GL hypomorphic variants in HGFR could be another, more subtle, genetic driver for maladaptive SGR. To test this hypothesis, we stringently collated pathogenically suspicious GL variants of several HGFR in a large cohort (n=7197) of patients with MN and or BMF and identified 59 BMF and 78 MN cases suspected to have GL alterations in one of the HGFRs (CSF1R, CSF2RB, CSF3R, MPL, and KIT). We subsequently analyzed these cases through the course of disease evolution until eventual diagnosis with MN or stability of a diagnosis of aplastic anemia (AA) or single- or multi-lineage cytopenias (SLC,MLC).

Of 46 patients with GL CSF3R variants, diagnoses were MN in 28 (17 AML, 6 MDS, 3 MDS/MPN, 2 MPN), AA in 3, MLC/SLC in 15. Median age was 64 and 55yr for MN and BMF, respectively. Of patients who developed MN (median time to evolve 40 m), 10/28 had prodromal cytopenia that lasted >6mo and median time to MN was 40 mo. Median OS for MN was 22mo. A patient presented with CNL and progressed to sAML in 21mo. Compared to general population, CSF3R GL variants resulted in increased total allelic burden. Most common somatic mutations in MN were FLT3 (8/28), ASXL1 (7/28), not observed in BMF patients (p=.015, p=.032). In 4/28 MN cases with somatic CSF3R^R583H (3) and CSF3R^T618I mutations, GL CSF3R^T640Iand CSF3R^G731R were detected. We also found 23 patients with GL KIT alterations, 12 with MN, 9 with MLC/SLC and 2 with AA. Median age was 66 for MN, 70 for BMF. Of MN progressors (median time of 52mo), 5/12 had cytopenias prior to neoplastic evolution, which was typically characterized by the acquisition of SRSF2 mutations (5/11). Of 18 patients with GL CSF1R variants, diagnoses were MN (9), AA (3), MLC/SLC (6). Median age of MN patients was 59 yr; 5/9 of MN had cytopenia >6mo. preceding MN. and median time to evolution was 25mo. Median OS was 7mo for MN. Median age of MN and BMF patients were 64 and 29yr (p=.037). Most common somatic mutations in MN patients were MPL and NPM1 (3 each), not observed in BMF patients. There were also 18 patients with GL MPL variants, diagnoses were MN (15), MLC/SLC (3). IDH2 (3) was most common in MN, and BMF patients had no somatic mutations detected. Of 17 patients with GL CSF2RB variants, diagnoses were AA (7), SLC/MLC (10). Median OS was 112mo. Most common mutations were CSF2RB and STAT3 (n=2 each). There were 6 patients with PNH clones, 4 with GL CSF2RB and 2 with GL CSF1R variants. Overall, 21 carriers of GL HGFR variants had somatic GOF in HGFR genes in biallelic or compound heterozygous (CH) configuration. Diagnoses were MN (17), AA (2), MLC (2). Of all MN, 5 presented with cytopenia before progression, median time to evolution was 75mo. Most common biallelic variant was CSF3R/CSF3R (4), 3 presented with AML and 1 presented with bicytopenia and progressed to MDS-SLD after 84mo. Most common CH combination was CSF3R^GL/FLT3^SM (9), 3 presented with cytopenias before progressing to AML at a median time of 75mo. Rest of the 6 patients presented with AML. In BMF, most common biallelic variant was CSF2RB/CSF2RB (2) both in AA patients and most common CH was CSF3R^GL/CSF2RB^SM (1) in a patient with neutropenia and thrombocytopenia. Of interest, 0.04% of screened samples were GL FLT3 variants (FLT3^V317E, FLT3^Q771P, FLT3^D62G). Diagnoses were AA, AML, MLC. AML evolved in the setting of RUNX1 and DNMT3A mutations, while patients with BMF had CSF2RB mutations. In summary, our study has shown that a subgroup of MN patients had an uncharacteristically long prodromal BMF phase before disease evolution. Some of these were characterized by somatic GOF mutation in HGFR, suggesting GL alterations may have played a role in selective pressure for evolution of disease. Another implication of our findings that supported this hypothesis was the non-random distribution of somatic mutations.