Mass Cytometry Analysis Of Myelofibrosis and Secondary Acute Myeloid Leukemia Reveals Constitutive and Cytokine Induced Signaling Abnormalities With Differential Sensitivities To Ruxolitinib

The classic myeloproliferative neoplasms (MPNs), including primary and secondary myelofibrosis (MF), are frequently associated with the JAK2 V617F mutation or other genetic alterations in members of the JAK-STAT axis. These mutations induce hyperactivated JAK-STAT signaling in cell lines and mouse models. Other mutations have been found recurrently in human MPNs, however, including mutations associated with transformation of chronic MPNs to secondary acute myeloid leukemia (sAML). The aggregate effects of these mutations on in vivo myeloproliferative signaling and disease phenotypes are not yet well understood.

While targeted inhibitors of JAK2 have shown activity in MPNs, evidence of a selective effect on the underlying malignant clone has been lacking. These findings suggest that dysregulation of other signaling molecules may be important in MPN pathogenesis, and that therapeutic targeting of these molecules could be beneficial. Therefore, a more complete assessment of JAK-STAT and related signaling pathways in MPNs is needed.

Mass cytometry is a novel technology that merges flow cytometry with mass spectrometry and enables the simultaneous measurement of 30+ parameters at the single cell level. We utilized this approach to examine bone marrow or peripheral blood samples from four MF patients and five sAML patients, as well as three normal controls. Cells were exposed to five different perturbation conditions ex vivo, including the cytokines thrombopoietin (TPO) and G-CSF, and/or the JAK1/2 inhibitor ruxolitinib. Cells were stained with a panel of 18 surface markers and 16 intracellular signaling effectors and analyzed on a CyTOF mass cytometer.

Single cell data was analyzed using traditional gating strategies, as well as with SPADE (spanning-tree analysis of density-normalized events), which distills multidimensional data down to interconnected cell subsets based on shared surface marker expression. These methods grouped cells into distinct cell populations including hematopoietic stem/progenitors (HSPCs) and myeloid and lymphoid lineage subsets. Heat maps were constructed to depict basal and induced activation of each intracellular marker.

MF and sAML patient samples exhibited abnormal cytokine-induced signaling that varied in a patient-specific manner. HSPC responses to TPO ranged from hyposensitive to hypersensitive compared with normal. Markedly abnormal basal (unstimulated) signaling in the HSPC compartment was observed in all MF and sAML patients. Each sample exhibited increased total IkBα and/or pCREB, typically accompanied by elevated phosphorylation of one or more additional signaling molecules. Both basal elevation and cytokine hypersensitivity were frequently observed within a MAP kinase signaling axis represented by pERK and downstream targets pCREB and pS6. Conversely, repressed basal STAT1 phosphorylation was observed in all patients. These abnormalities were not exclusive to sAML versus MF, nor specific to patients with or without the JAK2 V617F mutation. Widespread dysregulation of total IkBα, pERK-pCREB-pS6, and pSTAT1 suggest that these signaling effectors may be characteristic of a general myeloproliferative signaling phenotype.

Ex vivo ruxolitinib effectively inhibited the majority of observed hypersensitive cytokine-stimulated signaling effects. In contrast, most basal signaling elevations were insensitive to ruxolitinib. The signaling molecules most frequently insensitive to ruxolitinib were those most frequently basally elevated in MF and sAML HSPC, namely pCREB and total IkBα. In contrast, phosphorylation of the JAK2 substrates STAT3 and STAT5 was almost always sensitive to ruxolitinib. These results suggest that targeting of myeloproliferative signaling pathways either downstream or independent of JAK2 activity may be valuable for improved treatment of MF and sAML patients.

Ongoing experiments are focused on determining whether constitutive ruxolitinib-insensitive signaling abnormalities can be identified in patients treated with ruxolitinib in vivo, which may underlie the incomplete responses observed clinically. Taken together, these approaches will provide a deeper understanding of altered signaling networks in the context of targeted therapies in MPNs.