CBL Variants in Chronic Myelomonocytic Leukemia Exhibit a Complex Sub-Clone Architecture with a High Frequency of RING Domain Mutations Sufficient to Induce GM-CSF Hypersensitivity

Introduction: Chronic myelomonocytic leukemia (CMML) cells are hypersensitive to the pro-inflammatory cytokine granulocyte-macrophage colony-stimulating factor (GM-CSF). CBL mutations are frequently detected in CMML but molecular features, biology and optimal management is not known. Here we examined the clinical, molecular and immunophenotypic features of CMML patients with CBL mutations enrolled in the ongoing PREcision Approach to Chronic Myelomonocytic Leukaemia (ACTRN12621000223831) trial which assesses the efficacy of Lenzilumab (LENZ; Taran Therapeutics, NJ) with azacytidine (AZA) in CMML patients harbouring RAS pathway mutations (KRAS, NRAS, CBL) at a variant allele frequency (VAF) > 3%.

Method: VAF was determined from bone marrow (BM) obtained at baseline and day 1 of treatment cycles 4, 7, and 13, using a 46-myeloid gene panel analyzed by next generation sequencing (Illumina NextSeq). MNCs were immunophenotyped using a CD45, 14, 16, 34, 115, 116 and 131 antibody panel. Cord-blood derived CD34⁺ cells were nucleofected (Lonza) with CBL-directed RNA guides (Synthego, USA) and assayed for colony formation. Student’s t-test, Mann-Whitney and Fisher’s exact test were used for statistical analyses.

Results: Samples from 24 subjects were included for analysis. CBL mutations were detected in 46% (11/24) of subjects with two or more CBL variants detected in 64% (7/11) of subjects, in contrast to NRAS (40% with 2 of more NRAS variants, 2/5) and KRAS (38% with 2 or more KRAS, 3/8). Subjects with CBL mutations had increased BM blast percentage (10.1 ± 1.5 vs. 5.3 ± 1.4%, P=0.045), white cell count (26.8 ± 5.9 vs. 13.8 ± 5.2×10⁹/L, P=0.046) and splenomegaly (10/11, 91%, P=0.03) compared to subjects wildtype for RAS pathway, consistent with a more proliferative phenotype. 36% of cases with CBL mutations were classified as CMML-2.

Merging data from our cohort with CMML cases reported in COSMIC, we found a majority of mutations clustered within the RING domain of Cbl (94/137; 69%). This was in contrast to juvenile myelomonocytic leukemia (JMML), where linker domain mutations were most common (27/46; 59%) (P<0.0001), even though the incidence of CBL mutation is similar for the two diseases.

Immunophenotyping of MNCs, including those with CBL mutations, revealed a higher percentage (89.7 ± 1.6 vs. 50.3 ± 2.7%, P=10^-6) and density (40.2 ± 7.4×10³ vs. 13.9 ± 4.1×10³, P=0.08) of CD116 (GM-CSF receptor subunit a)-expressing CD34⁺ progenitors compared to healthy controls, coupled with a higher percentage of CD131 (beta common receptor) (64.3 ± 3.8 vs. 32.1 ± 4.1%, P=0.001). In contrast, CD115 (macrophage-colony stimulating factor receptor) was expressed at a lower density (14.6 ± 1.6×10³ vs. 29.5 ± 5.6×10³, P=0.005) in progenitors compared to controls.

Interestingly, CRISPR targeting of CBL with sgRNAs to the Linker and RING domain in human CD34⁺ cells (n=2 independent donors) resulted in increased short-term proliferative capacity in the absence of GM-CSF (fold change 3.02 ± 0.37; P=0.007) compared to control. Additionally, CRISPR-edited CBL mutant CD34⁺ cells phenocopied GM-CSF hypersensitivity in vitro, displaying increased colony formation in low concentration GM-CSF (1 ng/mL) (73.2 ± 6.5 colonies vs. control 16.5 ± 5.2, P=0.01) that was suppressed by LENZ (39.0 ± 3.6 vs. control 79.7 ± 9.7, P=0.002).

Notably, one subject with a CBL linker domain mutation that was sensitive to LENZ in colony assays (6.0 ± 2.7 vs. control 31.7 ± 12.5, P=0.02), showed a corresponding decrease in CBL VAF occurring within 3 months of LENZ/AZA therapy. Serial VAF analysis of 10 CBL variants from 7 subjects treated with LENZ/AZA revealed decreases in 9 of 10 of variants, particularly in clones driven by linker domain mutations, with some durable beyond 6 months (ADE03 VAF 75%, 42%, 27%, 11%; MEL05 48%, 6%, 8%; BRI07 55%, 0.7%; time points evaluated at screening, 3, 6 and 12 months respectively).

Conclusion: CRISPR engineering of CBL mutations in CD34⁺ progenitors was sufficient to induce GM-CSF hypersensitivity. CBL mutation confers a strong proliferative phenotype in CMML with many patients harboring multiple CBL subclones, especially in the RING domain. A high proportion of CD116-expressing progenitors, rather than CD115, indicates they may be targetable by anti-GM-CSF neutralization. The sensitivity of CBL mutants to GM-CSF blockade has implications for the treatment of CMML and JMML.