Abstract Background: Despite the identification of a number of genes that underlie the genetic complexity of the myelofibrosis (MF) phenotype and even with the subsequent FDA approval of a targeted therapy for MF that is ruxolitinib, the median overall survival time of MF patients is still only 5-6 years. This unchanged survival may be explained, in part, via the complex mutation patterns that contribute to MF. Consequently, treating MF patients remains a significant challenge in the hematology clinic. Aim: To identify new drugs for MF patients, we used predictive simulation modeling of JAK2-V617F expressing MPN cells that could readily be re-purposed for MF. Methods: : To identify combinations of other drugs that can enhance the effect of Ruxolitinib, we (1) employed predictive simulation to model a JAK2-V617F bearing cell line (2) identified the key molecular disease characteristics of the JAK2-driven disease network, and (3) experimentally validated the predictions. The predictive simulation approach from Cellworks provides a representation of disease physiology incorporating signaling and metabolic networks with an integrated phenotype view. We modelled the JAK2-V617F expressing SET2 cell line based on its genomic aberration data. We then screened a library of drugs on the computational model in combination with Ruxolotinib, by assessing the impact on disease specific biomarkers and phenotypes of proliferation, viability, and apoptosis. Results: The modeling and digital drug screening identified a short list of drugs that were predicted to inhibit Jak2 signaling. Validation studies were conducted with the representative JAK2 driven SET2 cell line which harbors the JAK2-V617F mutation and was predicted to lead to subsequent activation of STAT3, STAT5, ERK via SHC1, AKT via IRS1-PIK3CA and STAT1 (Fig 1). JAK2 is known to phosphorylate STAT3 on Y705. Predictive modeling also discovered activation of parallel signaling via mutations of ALK, IRS2 and RPTOR. ALK is also known to activate STAT3 via Y705 phosphorylation and ERK via SHC1. IRS2 leads to activation of PIK3CA-AKT-MTOR. SET2 cells also harbor a RPTOR mutation which also activates MTOR pathway further. With higher ERK and MTOR signaling, there was S727 phosphorylation of STAT3. Using digital drug screening on the SET2 simulation model, sorafenib mitigated the effect of JAK2 mutation by inhibiting pathways downstream of JAK2, rather than decreasing the JAK2 mutation burden itself. Sorafenib is a small molecule kinase inhibitor that targets multiple proteins including VEGFR, PDGFR, and several Raf family kinases. It is approved for use in renal, liver, and thyroid cancers and is under clinical investigation for a number of other neoplasias. A potential role for sorafenib in MF has not been fully investigated. Sorafenib is known to inhibit ERK via inhibition of RAF. Sorafenib was also predicted to inhibit STAT3, STAT5, STAT1 and AKT1 via activation of PTPN6 in the SET2 model. With down-regulation of ERK and AKT1-MTOR pathway, Sorafenib was predicted to inhibit STAT3 S727 phosphorylation. Hence, with sorafenib and ruxolitinib, not only was JAK2 burden reduced, but also parallel pathway signaling via ALK, IRS2 and RPTOR. Using in vitro validation, we found that sorafenib was highly efficacious in reducing Jak2-V617F cell viability in a dose-dependent manner (GI50 = 6 uM) . Moreover, when Jak2-V617F cells were treated with both ruxolitinib and sorafenib, there was synergy, suggesting that the compounds were acting on multiple, cooperating signaling pathways. A prior report also showed synergistic MPN regression with ruxolitinib and sorafenib in BaF3/Jak2-V617F and HEL cell lines and primary specimens (PMID: 23560534), thereby providing independent evidence for the rationale use of these compounds in MF patients harboring mutation of PDGFRA/B, NRAS, KRAS in addition to JAK2 mutations. Conclusion: Thus, in addition to identifying a novel drug regimen for MF, we demonstrate that successful treatment of MF likely requires targeting multiple, non-redundant, constitutively active, and cooperative signaling pathways, rather than a one-biomarker/one-drug approach as is the current clinical standard. Further, we demonstrate how predictive simulation modeling can be used for precision medicine purposes in MF to inform clinical treatment decision-making of FDA approved drugs, with the intent of providing significant health and cost benefits. Download : Download high-res image (220KB) Download : Download full-size image Figure 1 . Disclosures Singh: Cellworks Research India Pvt. Ltd: Employment. Pampana: CellWorks: Employment. Usmani: Cellworks: Employment. Sauban: Cellworks: Employment. Kushwaha: CellWorks: Employment. Ramakrishnan: CellWorks: Employment. Abbasi: Cellworks Group Inc.: Employment. Vali: Cellworks Group Inc.: Employment. Cogle: Celgene: Other: Membership on Steering Committee for Connect MDS/AML Registry.
Read full abstract