Molecular Spectrum of CSF3R variants Correlate with Specific Myeloid Malignancies and Secondary Mutations

Different CSF3R mutations (CSF3R^MT) result in aberrant G-CSF signaling pathways and are linked to a wide range of myeloid disorders. Loss-of-function mutations in its extracellular domain cause severe congenital neutropenia (SCN). Activating mutations in the juxtamembrane region have been associated with a variety of myeloid malignancies. Truncating mutations in the cytoplasmic domain are associated with SCN cases that progress to MDS or AML. In this study, we evaluate the extent to which different CSF3R^MT associate with disease onset, progression to leukemia and neutrophil counts in patients (pts) diagnosed with myeloid malignancies.

We identified CSF3R^MT cases in a cohort of 1400 pts [median age 71 years (yrs)]. We analyzed somatic and germline mutational patterns, and cross-sectional correlation with other gene mutations in CSF3R^MT. A stringent algorithm based on conserved amino acid residues and alterations of protein features was used to predict the pathogenic significance of CSF3R^MT. We identified 44 CSF3R^MT: 33 germline (CSF3R^GL) and 11 somatic (CSF3R^S) variants.

Most CSF3R^GL were found in pts (median age 63 yrs) with MDS or related conditions (87% of all mutant cases), conversely these mutations were present in 5% (n= 22/424) of MDS, 3% (n= 7/244) MDS/MPN and <1% (n= 3/392) of AML and in 1 out of 3 pts with aCML tested. Mutations were mostly missense and located between the cytoplasmic (58%: M696T, R698C (isoform III), D732N, P733T, S744F, Y752*, E808K), and extracellular (42%: C131Y, E149Q, A208V, Q216H, D320N, E405K, S413L, Y562H) domains. No mutations were detected in the juxtamembrane domain. Variants were grouped in Tier-1 (61%: C131Y, E149Q, A208V, Q216H, D320N, E405K, S413L, Y562H Y752*, E808K) and Tier-2 (variants with uncertain significance, 39%: S413L, M696T, R689C, D732N, P733T, S744F). E808K and R698C were the most common amino acid changes in Tier-1 (53%) and Tier-2 (44%), respectively. A total of 4/7 pts with E808K progressed to AML (but none with R698C), supporting previous observations that E808K (or E785K) represents a pathogenic variant predisposing to leukemia. A total of 46% (n=14) of pts with CSF3R^GL had neutropenia [median 0.9x10⁹/L (0.02-1.22x10⁹/L)] at the time of sampling. Two pts diagnosed with a prior cancer manifested sustained neutropenia before the diagnosis of MDS and MDS/MPN. G-CSF was administered in 21% of pts. Alterations in -7/7q- were common (21%). Some pts also harbored other somatic mutations in NF1 (15%), DNMT3A (12%), SETBP1 (12%), or U2AF1 (12%). Of note, 1 patient carried mutations in WAS and GATA2 and another carried a mutation in VPS45, which have been previously associated with SCN/MDS. The patient with aCML harbored also a CSF3R^S (T615A).

Overall combined allelic burden in pts cohort was 2% vs. 1.6% expected allelic burden in control populations for the same variants (P=.02).

CSF3R^S were found in 11 pts (median age 71 yrs) with AML or MDS related conditions (73% of all mutant cases), conversely these mutations were present in 1.4% (n= 6/424) of AML, <1% in MDS (n= 2/244) and MDS/MPN (n= 1/392) and in 2/3 pts with aCML tested. Mutations were missense in 63% of pts, T618I being most recurrent (n=5/11; 45%). Frameshifts accounted for 36% of the mutations and were localized in the cytoplasmic domain (Q741*, Q749*, Y752*, Q768*). All mutations were heterozygous. At the time of sampling 3/11 pts had leukocytosis and 3/11 had neutropenia. Mutations were distributed between the juxtamembrane domain (55%) and the cytoplasmic domain (45%). Mutations in the extracellular domain were not detected. Pts with sAML mostly carried mutations in the juxtamembrane domain (67%), those with MDS carried only in cytoplasmic domain, and those with MDS/MPN or aCML carried mutations in both the juxtamembrane and extracellular domains. There was one somatic and one RUNX1^GL mutation. The cytogenetic abnormalities -7/7q- were detected in 18% (2/11) of cases. Interestingly, T618I was found solely in pts with sAML.

Focusing on associations between CSF3R^MT and mutations in the class III receptor tyrosine kinases CSF1R, FLT3, and KIT we identified only FLT3 to be co-mutated with CSF3R^MT. All 3 pts (2 CSF3R^GL and 1 CSF3R^S) with such co-mutations evolved to AML.

In sum, we found that CSF3R^GL do not commonly co-occur with CSF3R^S, suggesting that the neutropenia observed at the sampling time most likely is causative of undetected GL variants and/or is representative of a long unrecognized disease.