Aberrant Cytokine Production in Myelofibrosis Is Not Rectified By Ruxolitinib and Is Differentially Sensitive to Inhibition of JAK/STAT, MAP Kinase, and NFκB Signaling

Inflammatory cytokine production is characteristic of myeloproliferative neoplasms including myelofibrosis (MF). Clonal dominance and bone marrow fibrosis have been attributed in part to cytokine overproduction. Elevated plasma levels of several cytokines have been associated with decreased survival in MF patients; notably IL-8/CXCL8, which has been associated with transformation to secondary acute myeloid leukemia (sAML; Tefferi et al., 2011, J. Clin. Oncol.).

In MF patients, JAK2 inhibition with ruxolitinib ameliorates constitutional symptoms and splenomegaly; but improvements in anemia, fibrosis, and malignant clonal burden are infrequent, and survival benefit is modest. We have measured blood plasma cytokines in MF patients prior to and on treatment with ruxolitinib, versus healthy controls and sAML patients. Cytokines significantly elevated in MF patients versus controls (VEGF, TNF, IL-6, IL-10, IL-16) showed minimal differences in pre-versus-on ruxolitinib levels. Indeed, both pre- and on ruxolitinib cytokine levels in MF were similar to those observed in sAML. Therefore, overall cytokine-producing activity in MF is of similar magnitude to that in leukemic hematopoiesis, and persists despite ruxolitinib treatment.

To identify cell populations responsible for cytokine production, MF blood cells were studied using mass cytometry (CyTOF). In MF versus healthy control samples, 14/15 cytokines measured were found to be constitutively overproduced in myeloid cell populations. IL-8 was highly expressed in CD34+ hematopoietic stem and progenitor cells (HSPC), which are prevalent in MF blood. The prognostic impact of IL-8 may therefore reflect overproduction of phenotypic HSPC, which could indicate incipient transformation to sAML. For the majority of MF overproduced cytokines, however, the largest source population was monocytes. This indicates that cytokines produced by monocytes may impact disease-propagating HSPC in a non-cell-autonomous fashion.

To evaluate mechanisms driving cytokine overproduction, MF patient and control samples were exposed to 4-hour incubations with TPO, TNF, and/or TLR ligands PAM3CSK4 and R848. These compounds activate either JAK/STAT or NFκB signaling. The majority of cytokines elevated in MF exhibited hypersensitivity to these agents in MF versus normal monocytes. TPO induced cytokines in monocytes via its receptor MPL, which was expressed in a subset of both normal and MF monocytes. Cytokine induction by TPO was observed in normal and MF monocytes isolated by flow sorting (CD14+CD34-CD61-), confirming that MPL and JAK2 can induce cytokine overproduction in monocytes cell-autonomously. TPO induced cytokine production in monocytes could be suppressed either by ruxolitinib or pevonedistat, a NEDD8-activating enzyme inhibitor that blocks NFκB activation. This implies that both JAK2 and NFκB signaling pathways must be intact for maximal cytokine induction downstream of MPL.

In contrast to TPO induced cytokine production, basal constitutively elevated cytokine production in MF monocytes was not consistently reduced by ruxolitinib. Among the TPO/TNF/TLR inducible cytokines, a subset (including TNF, IL-6, IL-8, IL-10) was minimally sensitive to ruxolitinib. All TPO/TNF/TLR-sensitive cytokines, however, were at least partially sensitive to pharmacologic inhibition of NFκB and/or MAP kinase signaling, as observed from incubation with pevonedistat, trametinib (MEK inhibitor), VX-745 (p38 MAPK inhibitor), and/or JNKi8 (JNK inhibitor). In contrast, these inhibitors did not suppress basal production of non-TPO-inducible cytokines (e.g. TGF-β, IFNγ), which are also overproduced in MF.

These results indicate that overproduction of one set of cytokines in MF is driven by a combination of JAK/STAT, NFκB, and MAP kinase signaling; while that of another set is not directly driven by any of these pathways. Cytokines whose production is refractory to ruxolitinib may underlie features of MF that resist ruxolitinib therapy. A combined inhibitory approach may provide therapeutic benefit beyond that observed with ruxolitinib. Based on this hypothesis, our group has initiated a Phase 1 clinical trial combining pevonedistat with ruxolitinib for MF treatment (NCT03386214). Further results will indicate whether inhibition of NFκB confers an additional benefit to that observed from ruxolitinib alone.