Comprehensive Cytokine Profiling of Patients with Advanced Systemic Mastocytosis Treated with Midostaurin

Background: Advanced systemic mastocytosis (AdvSM) comprises a group of hematologic neoplasms primarily driven by the oncogenic KIT D816V mutation. Morbidity and mortality relate to organ damage caused by neoplastic mast cell infiltration and mast cell mediator symptoms. Midostaurin, an oral multikinase/KIT inhibitor, produced reversion of organ damage in 60-70% of AdvSM patients (Gotlib et al, NEJM, 2016; DeAngelo et al, Leukemia, 2018), which was associated with decreases in bone marrow mast cell burden, serum tryptase levels, and splenomegaly in the majority of patients. Similar to myelofibrosis (MF) patients treated with JAK inhibitors, midostaurin improves disease symptoms and quality of life (Gotlib et al, NEJM, 2016). We conducted a cytokine profiling study to better understand the biologic effects of midostaurin in this patient population.

Methods: Plasma cytokine levels were analyzed in 19 patients with AdvSM (2 aggressive SM, 10 SM with an associated hematologic neoplasm, and 7 with mast cell leukemia) treated with midostaurin 100 mg twice daily. Seventeen patients were treated on the phase II investigator-initiated trial (DeAngelo et al,Leukemia, 2018), and two were treated on a compassionate-use basis. Cytokine levels were compared to 10 patients with MF (primary MF [n=4]; post PV/ET MF [n=6]), treated with JAK inhibitors (ruxolitinib, fedratinib, or momelotinib) and 10 healthy patients whose age range matched the SM group. Plasma samples were obtained pretreatment, and after 4-8 weeks of therapy. Each patient sample was evaluated for 62 cytokines using a Luminex assay (eBioscience) performed at the Stanford Human Immune Monitoring Center. Sparse partial least squares discriminant analysis (SPLS-DA) was used to identify the most predictive cytokines from the expression data that help classify patient samples based on disease states, and SPLS regression was used to identify cytokines correlated with survival (Rohart et al, mixOmics. R package version 6.3.2). Overall survival was calculated from the time of initiation of midostaurin to time of death.

Results: Plasma levels of several cytokines were higher at baseline in AdvSM and MF patients compared to healthy controls (Fig 1A-B). In patients with AdvSM and MF, pretreatment levels of leptin (false discovery rate (FDR) < 0.002), epidermal growth factor (EGF; FDR < 2e-5), and brain-derived neurotrophic factor (BDNF; FDR < 3.5e-7) were lower than healthy controls. Using SPLS-DA, we identified a distinct baseline cytokine expression profile that could classify patient samples as either AdvSM, MF, or healthy with an error rate of less than 0.10 (Fig 1C). The six most predictive cytokines were nerve growth factor, EGF, BDNF, interferon alpha, intercellular adhesion molecule-1 (ICAM1), and leukemia inhibitory factor. Corroborating prior data (Verstovsek et al, NEJM 2010), we found that several plasma cytokine levels significantly decreased in MF patients after 1-2 cycles of JAK inhibition (Fig 1B). Although we observed similar decreases in plasma cytokine levels for AdvSM patients on midostaurin (Fig 1A), the results were more variable and did not reach statistical significance. AdvSM patients that responded to midostaurin (major or partial response per Valent criteria) expressed a different baseline cytokine expression profile when compared to non-responders. SPLS-DA was able to distinguish these two classes of patients with an error rate of 0.35 (Fig 1D). SPLS regression identified baseline levels of interleukin-7 (IL7), ICAM1, interleukin 12 subunit p40 (IL12p40), EGF, and platelet-derived growth factor BB (PDGFBB) as potential predictors of survival for patients with AdvSM treated with midostaurin. High pretreatment levels of IL7 (109 months (m) vs. 33 m; p=0.01), EGF (92 m vs. 34 m; p=0.04), and PDGFBB (80 m vs. 37 m; p=0.09) and low pretreatment levels of ICAM1 (91 m vs. 19 m; p=0.02) and IL12p40 (91 m vs. 19 m; p=0.03) were associated with increased survival.

Conclusions: Cytokine expression profiling can distinguish patients with AdvSM or MF from healthy controls. Compared to the use of JAK inhibitors in MF, midostaurin's activity in AdvSM appears less related to reversion of pro-inflammatory cytokines. Investigation of a larger cohort of patients is needed to determine whether such analyses of plasma cytokines will provide added value to new scoring prognostic systems of AdvSM that incorporate clinical and molecular data.

View largeDownload slide

View largeDownload slide

Close modal