Correlation between interleukin (IL)-8 and tumor necrosis factor (TNF)-alpha levels and overall survival in patients (pts) with myelofibrosis (MF) relapsed or refractory (R/R) to a Janus-associated kinase inhibitor (JAKi) treated with imetelstat (IME) in the imbark trial

Abstract

Abstract Introduction: The dysregulation of inflammatory cytokines plays a key role in the pathogenesis, progression, and clinical features of MF. In particular, increased IL-8 and TNF-alpha have been associated with poor prognosis in pts with MF. In preclinical evaluations, the competitive inhibition of telomerase activity with IME led to decreased expression of IL-8 in various tumor cell lines. In the Phase 2 IMbark trial (NCT02426086) of 2 dosing regimens, IME showed a dose-related trend for clinical benefit, an acceptable safety profile, and potential disease-modifying activity in pts with intermediate-2 or high-risk MF R/R to JAKi treatment (tx). This post hoc analysis evaluates the effect of IME on pro-inflammatory cytokines and its impact on efficacy in pts from IMbark. Methods: In IMbark, pts were randomly assigned (1:1) to 4.4 mg/kg or 8.9 mg/kg IME active dose (equivalent to 4.7 mg/kg or 9.4 mg/kg IME sodium, respectively) infusion every 3 weeks. In this analysis, 21 cytokines were measured using Luminex® Multiplex Cytokine Assay Kits (R&D Systems, Inc.) in stored plasma samples collected at baseline (BL) and at week 12, 24, 36, and 48 of IME tx. The correlation between percent change from BL in cytokine levels and percent change in total symptom score (TSS) or spleen volume at each time point was evaluated using linear regression. Overall survival (OS) by BL cytokine level with IME (any dose or by dose level) was estimated using Kaplan-Meier; hazard ratio [HR] between cytokine levels was evaluated using Cox regression. An unadjusted nominal P value from each test was reported. Results: A total of 43 out of 107 pts from IMbark who had plasma samples available for cytokine evaluation and quantifiable cytokine levels at BL and ≥1 post-BL assessments were included in this analysis. Of 21 cytokines analyzed, a consistent trend of dose-dependent reduction across time points was seen selectively with IL-8 and TNF-alpha levels. The reduction from BL in IL-8 levels with IME (regardless of dose) at week 12, 24, and 36 was significantly correlated with reductions in TSS at the corresponding time points (correlation coefficients [CC] of 0.361 [P=.019], 0.319 [P=.045], and 0.349 [P=.040], respectively) and with spleen volume reductions (SVR) at week 12 (CC, 0.399; P=.009). IL-8 correlations with SVR at weeks 24 and 36 were not statistically significant. The reduction from BL in TNF-alpha levels with IME (regardless of dose) at week 12, 24, 36, and 48 was significantly correlated with SVR at the corresponding time points (CCs of 0.636 [P<.001], 0.457 [P=.003], 0.439 [P=.011], and 0.475 [P=.012], respectively). Correlations of TNF-alpha and TSS were not statistically significant. Pts who achieved spleen response (≥35% SVR) at 24 weeks with IME had mean reductions of 32% and 46%, respectively, for IL-8 and TNF-alpha levels at week 24; SVR nonresponders had mean reductions of 2% and 13%. Overall, pts with lower BL IL-8 levels (defined as ≤median level) had longer median (95% CI) OS with IME compared with pts with higher BL IL-8 levels (defined as >median level: HR [95% CI], 0.44 [0.21-0.93]; P=.03). Among pts with higher BL IL-8 levels, median (95% CI) OS was longer with 8.9 mg/kg IME (24.1 mo [18.4-31.6]) compared with 4.4 mg/kg (18.2 mo [14.7-not estimable]; HR [95% CI], 0.45 [0.14-1.48]; P=.18). Similar trends in OS were observed for TNF-alpha, but these were not statistically significant: pts with lower BL TNF-alpha levels had longer median OS compared with pts with higher BL TNF-alpha levels (HR [95% CI], 0.64 [0.31-1.31]; P=.21). Among pts with higher BL TNF-alpha levels, median OS (95% CI) was longer with 8.9 mg/kg IME (22.8 mo [13.9-not estimable]) compared with 4.4 mg/kg IME (19.3 mo [14.7-34.7]; HR [95% CI], 0.68 [0.24-1.88]; P=.45). Conclusions: In this post hoc analysis, IME tx resulted in dose-dependent reductions from BL of IL-8 and TNF-alpha levels, which corresponded with reductions in TSS or spleen volume. Furthermore, 8.9 mg/kg IME showed a longer survival benefit compared with 4.4 mg/kg IME in pts with higher BL IL-8 and TNF-alpha levels who face worse prognosis, although pt numbers were small. This hypothesis-generating, exploratory analysis identifies specific inflammatory cytokines reduced with IME tx and associated reduction in spleen volume and symptoms that, together with previously presented data, suggest potential disease-modifying activity in MF. Further confirmatory studies of IME in MF are in progress.