Targeting JAK2 Gene Rearrangements with a Novel Kinase Inhibitor in a Preclinical Model of Pediatric Acute Lymphoblastic Leukemia

Background. Although the Event Free Survival for Childhood Acute lymphoblastic leukemia (ALL) reaches 85%, the remaining 15% of patients relapse, and 25-40% of them die. Novel molecular targets may increase the efficacy of therapy and reduce treatment toxicity. Among B-other ALL patients, JAK/STAT pathway alterations represent about 7% of the 'Philadelphia-like' cases. JAK2 gene encodes for a non-receptor tyrosine kinase fundamental for hematopoiesis, cellular proliferation and survival. In last years, JAK2 mutations have been widely studied in leukemia and lymphoma, whereas JAK2 fusion genes are still poorly characterized.

Aim. This study aims to identify JAK2 fusion genes among BCP-ALL pediatric patients and develop a target strategy in in vitro and in vivo preclinical models.

Methods. A targeted capture RNA Next Generation Sequencing strategy was applied to discover JAK2 fusion genes in a large cohort of PCR-based MRD high risk (HR) BCP-ALL pediatric patients. Fusions were validated by RT-PCR and/or FISH. Primary patients' cells have been in vivo expanded in NSG mice. We performed ex vivo and in vivo drug treatments with JAK2 inhibitors; phosphoflow and apoptosis-viability assays were performed in patients' blasts in co-culture with human bone marrow stroma.

Results. We identified 10 pediatric cases carrying a JAK2 fusion gene with different partners in single cases, such as ATF7IP, ZEB2, MPRIP, BCR, TLE4, GIT2 and RAB7, in addition to PAX5, which was the only recurrent in three cases. Cells were available from 3 cases, carrying PAX5-JAK2, ATF7IP-JAK2 and ZEB2-JAK2, respectively. After in vivo expansion, we demonstrated that the JAK2 signaling pathway was active at basal level, through phosphorylation on Y1007-1008 JAK2 residues inside the catalytic activation loop, compared to cases wild type for JAK2 and CRLF2 (+70%, two-tailed P value 0.0355); a positive trend was also shown compared to primary cells with P2RY8-CRLF2 rearrangements and JAK2 mutation, as positive controls (+40% two-tailed P value 0.158). The JAK2 downstream effectors pS727-STAT3 and pY694-STAT5 were also activated.

We thus setup a JAK2 targeted drug treatment using CHZ868, a new class-II tyrosine kinase inhibitor (TKI) (Novartis, Basel, CH). After 30 minutes of treatment, we appreciated a mean inhibition of -62% of Y1007-1008 JAK2 residues in PAX5-JAK2, -22% in ATF7IP-JAK2 and -35% in ZEB2- JAK2. Contemporarily, we observed a decrease of pS727-STAT3 (-35-50%) and pY694-STAT5 (-15-50%).

After 48h monotherapy treatment by CHZ868, we detected apoptosis induction and cell viability decrease between 20- 75% at IC50. In combination with dexamethasone, we assessed a further decrease of viability between 10 to 95%. A biological variability among the three different patients was appreciated, according to the different partner genes.

Exclusively for the PAX5-JAK2 fusion, we also performed treatments with the kinase inhibitor BIBF1120/Nintedanib, targeting LCK, which is activated downstream PAX5 fusions and we observed a 20% reduction of cell viability. Importantly, combination of BIBF1120 and CHZ868 showed a synergistic effect (-45%, at IC50). Moreover, we found that ruxolitinib caused autophagy as observed by higher levels of LC3-II compared to untreated cells (+ 45%, p<0.01), with consequent reduction of apoptosis induction. Indeed, active caspase 3 increased when ruxolitinib was given in combination with chloroquine, an autophagy inhibitor (+20% vs ruxolitinib alone, p<0.01). CHZ868 alone or in combination with chloroquine instead does not induce autophagy as LC3-II and active caspase 3 levels are the same of untreated cells.

Finally, we demonstrated the in vivo efficacy of CHZ868 in patient derived xenograft model in presence of PAX5-JAK2 fusion. After two weeks of 30mg/Kg daily treatment of CHZ868, we observed a significant reduction of leukemic CD10+/CD19+ cells both in bone marrow (p<0.01, -43%), spleen (p<0.001, -72%), central nervous system (-40%) and peripheral blood (p<0.05, -46%), compared to vehicle mice. Further in vivo experiments are ongoing in other JAK2 fusion settings.

Conclusion. CHZ868 is a promising candidate for treatment of BCP-ALL carrying JAK2 fusions, showing high efficacy and specificity, both ex vivo and in vivo. Further studies will include combination with standard chemotherapy drugs with the aim to maintain its efficacy by reducing the intensity and toxicity of chemotherapy.