Ruxolitinib-Induced Modulation of Bone Marrow Microenvironment in Patients with Myelofibrosis Is Associated with Inflammatory Cytokine Levels

Background: The Janus kinase inhibitor ruxolitinib (RUX) has recently been shown to induce a meaningful reduction in bone marrow (BM) fibrosis in patients with intermediate-2 or high-risk myelofibrosis (MF) compared with best available therapy (BAT). The altered BM microenvironment in MF is generally associated with prominent pro-inflammatory changes and high levels of corresponding cytokines.

We evaluated the complex BM stromal changes following up to 5 years of RUX therapy in a cohort of patients (pts) with primary, post-polycythemia vera, or post-essential thrombocythemia MF.

Methods: BM biopsies were performed at protocol-defined intervals for RUX-treated pts at baseline (BL) and at 24 months (mo) (68 pts), 48 mo (38 pts), 54 mo (23 pts), 60 mo (10 pts), or 66 mo (4 pts). To compare the effects of long-term therapy on BM fibrosis we additionally included a historical cohort of pts with primary MF treated with BAT. Samples from the BAT cohort (192 pts) were matched with those from the RUX cohort according to BM morphology at BL. Analysis in the RUX cohort included immunohistochemical assessment of therapy-induced changes by applying specific antibodies (CD61 – megakaryocytes [MEG]; CD68 – macrophages [MAC]; MUM1 – plasma cells [PLC]). Changes in these parameters were determined by consensus after independent blinded review by 3 pathologists (HMK/JT/CBR). Results were categorized as improvement, stabilization, or worsening from BL. Changes in serum levels of 73 cytokines (BL vs 4 weeks) were available in 48 (of the 68; 71%) pts of the RUX cohort. Logarithmic transformation was applied to both BL and week 4 cytokine values, and a stepwise logistic regression method was implemented to identify significant predictors. A composite variable was constructed that reflects a linear combination of the most important predictors using the slope from the logistic analysis as the coefficients. ROC curves with the status of BM response as the dependent variable were generated and the area under the curve (AUC) was calculated as a measure of accuracy.

Results: In line with previous results, the odds for improvement or stabilization of BM fibrosis were greater with RUX at all time points, whereas the odds of worsening were markedly reduced with RUX vs BAT. Improvement or stabilization of BM fibrosis with RUX was associated with concordant changes in both collagen deposition and osteosclerosis. As an indicator of therapy-induced reduction in the BM inflammatory stromal reaction, in more than 70% of pts a decrease in the amount of PLC was observed. In 48% of patients, RUX induced a reduction of MAC, while in 39% of cases their number remained unchanged. MEG frequency was reduced in 52% of cases post–RUX therapy, while an increase was observed in only 6.8% of samples. Furthermore, MEG atypia was reduced in 37% of cases, while in 55% of patients a stabilization effect of therapy was seen for this parameter. Multivariate regression identified 10 cytokines that strongly correlated with morphologic outcomes with different weighting factors for each of the predictive equations: CRP: C-reactive protein, IL-10: Interleukin 10, MDC: Macrophage- derived cytokine, SCF: Stem cell factor, TNFα: Tumor necrosis factor alpha, ApoA1: Apolipoprotein A1, EOT: Eotaxin, Hp: Haptoglobin, IgE: Immunoglobulin E, TIMP1: Tissue inhibitor of metalloproteinase-1. In all pts tested RUX induced a significant change of these identified cytokines by week 4. Individual changes in parameters of BM morphology were strongly associated with corresponding changes in cytokine levels. Decrease of BM fibrosis at 24 mo was strongly associated with reduced cytokine levels (AUC=0.85939, p=0.0134). Additionally, decreased frequencies of MEG, PLC, and MAC showed significant correlations with RUX-induced modulation of cytokines (MEG: AUC=0.942572, p=0.0176; PLC: AUC=0.838135, p=0.0331; MAC: AUC=0.965495, p=0.0033).

Conclusions: These results demonstrate that long-term RUX therapy, unlike BAT, causes a marked modulation of the BM cellular microenvironment. The effect observed with RUX was seen across all aspects of the fibrotic process including the inflammatory BM stromal reaction associated with MF. The complex relationships between improvement of the BM microenvironment and RUX-dependent effects on inflammatory cytokines reinforce previous observations suggesting that sustained JAK inhibition may be disease-modifying in MF.