Mass Cytometry Analysis Reveals Cellular Loci of Cytokine Overproduction in Myelofibrosis with Differential Sensitivity to Ruxolitinib

Inflammatory cytokine production is a characteristic feature of myeloproliferative neoplasms including myelofibrosis (MF). Inhibition of JAK2 with ruxolitinib ameliorates constitutional symptoms and splenomegaly in MF patients. However, improvements in anemia, fibrosis, and malignant clonal burden are infrequent. Therapeutic ruxolitinib has been shown to reduce, but not rectify, elevated plasma cytokine levels in MF (Verstovsek et al., 2010 N. Engl. J. Med.). Persistent inflammatory cytokine production may therefore contribute to the incomplete response to ruxolitinib observed clinically. The objective of this study is to define the aberrant cytokine-producing cell populations in MF and determine their response to ruxolitinib, with the goal of identifying strategies to manipulate these cytokines and their downstream signaling effectors for therapeutic benefit.

Our group recently reported the identification of hyperactivated NFκB signaling in MF, including elevated single-cell levels of phospho-p65/RELA, as observed by mass cytometry (CyTOF) (Fisher et al., 2017 Leukemia). NFκB activation was typically pan-hematopoietic in MF, suggesting a non-cell-autonomous etiology. Plasma TNFα levels were elevated in these patients, and mass cytometry analysis revealed that monocytes were the predominant source of TNFα. These findings suggest that excessive TNFα production could aberrantly activate NFκB across multiple cell populations in a non-cell-autonomous fashion.

To identify the cell populations that overproduce cytokines in MF, we analyzed cytokine levels with single cell resolution by mass cytometry. In MF versus healthy control blood and bone marrow samples, 14/15 cytokines measured were found to be constitutively overproduced in myeloid cell populations. The principal source cell population for the majority of MF elevated cytokines, including TNFα, was monocytes. In contrast, IL-8/CXCL8 was primarily produced by Lin-CD34+ cells, with consistently higher levels in MF versus healthy controls. Both monocytes and megakaryoblasts were observed to produce TGF-β, which has been implicated in bone marrow fibrosis. Patterns of cytokine overexpression, including source populations, were found to be consistent among JAK2 V617F (N=4), MPL W515L/K (N=2), and CALR (N=2) mutant MF patients.

To evaluate potential mechanisms driving cytokine overproduction, MF patient and control samples were exposed to 4-hour incubations with TPO, TNFα, and the TLR agonists R848 and PAM3CSK4. TNFα incubation for 4 hours was observed to induce expression of IL-6 and CCL4/MIP1β in MF monocytes, above basally elevated levels. Incubation with TPO or TLR agonists stimulated a greater number of cytokines, including TNFα in monocytes and IL-8/CXCL8 in Lin-CD34+ cells. A subset of MF elevated cytokines, including TGF-β, were contrarily not further induced by TPO or TLR agonists.

Cytokines overproduced in MF and inducible by TPO and TLR ligands could be divided into two groups on the basis of sensitivity to a 4-hour ex vivo incubation with ruxolitinib. One group of cytokines, namely IL-8, CCL3/MIP1α, and CCL4/MIP1β, showed reduction of unstimulated levels by ruxolitinib. In contrast, TNFα, IL-6, and IL-10, all mainly produced by monocytes, did not show reduction of unstimulated levels by ruxolitinib. An additional group of MF elevated cytokines, including TGF-β, did not exhibit either induction by TPO or TLR ligands, or suppression by ruxolitinib.

These findings indicate that in MF, individual cytokines are differentially produced by monocytes and other myeloid populations, and exhibit variable responses to ruxolitinib. The ruxolitinib-insensitive MF overproduced cytokines are likely to contribute to features of MF that are insensitive to ruxolitinib in vivo . Ongoing experiments are focused on identifying specific consequences of cytokine elevation in MF, as well as evaluating the dependence of cytokine overproduction in MF on specific signaling pathways. A goal of further study is to evaluate whether targeting MF elevated cytokines, their receptors and downstream signaling, that persist in the setting of ruxolitinib treatment, is likely to be alterable to produce therapeutic benefit.