The Dual PI3Kδγ Inhibitor Duvelisib Potently Inhibits IL-6 Production and Cytokine Release Syndrome (CRS) While Maintaining CAR-T Function in Vitro and In Vivo

Despite remarkable clinical efficacy, CAR-T therapy has been limited by life-threatening toxicities in over 30% of patients (Maude, NEJM 2014 and Davila, SciTransMed 2014). Toxicities primarily manifest as Cytokine Release Syndrome (CRS) characterized by an early phase with fever, hypotension, and elevations of cytokines including IFNγ, GM-CSF, TNF, IL-10, and IL-6. Using a protein kinase inhibitor library containing 644 independent compounds, we aimed to identify compounds that could block CRS-related cytokine production without inhibiting CAR-T function. We identified, duvelisib (kindly provided by Verastem Oncology, Needham, MA), a novel and selective dual PI3K-δ,γ inhibitor as a potent inhibitor of CRS in vitro and in vivo without attenuating CAR-T function. Duvelisib (Copiktra) is approved for the treatment of relapsed/refractory CLL after 2 prior therapies and follicular lymphoma after 2 prior systemic therapies; the latter gained accelerated approval status based on overall response rate and continued approval may be contingent on confirmatory trials. To assess the ability of duvelisib to inhibit CAR-T mediated CRS, we performed an in vitro CRS assay (Singh, Cytotherapy, 2017). CART19 (19-28BBζ, 25,000 cells), Ramos (CD19+, 50,000 cells), and immature dendritic cells (iDC, 2,500 cells) were co-cultured in a 96 well plate for 48hrs in the presence of varying concentrations of duvelisib (0.3nM-1000nM). Secreted IL-6, a surrogate marker of CRS, was determined using a human IL-6 ELISA (R&D Systems). Duvelisib reduced IL-6 levels in a dose-dependent manner with 30nM duvelisib reducing IL-6 secretion more than 10-fold (Fig 1a). To confirm that duvelisib did not inhibit CAR-T function, we performed in vitro killing assays, in which CAR-T efficacy was determined using BLI imaging of luciferase labeled CD19+ Ramos targets. At clinically relevant therapeutic doses (C max of 200 to 500nM) of duvelisib, there was no effect on CAR-T function in vitro. Treatment with 10nM duvelisib resulted in a statistically insignificant ~20% reduction of CART19 efficacy (p>0.05) (Fig 1b). Of interest, although selective inhibitors of either or PI3Kδ (GSK2292767) or PI3Kγ (IPI549) had modest effects on blocking CAR-T induced IL-6 production in this in vitro model, the effect of combining both inhibitors had a more dramatic effect similar to the dual PI3K-δ,γ inhibitor, duvelisib. Next we assessed the ability of duvelisib to block IL-6 secretion in vivo using a fully immunocompetent murine model of CRS. Six-week old BALB/c mice were injected with the mitogenic anti-CD3ε antibody, 145-2C11 (10 µg/mouse). Duvelisib (450µg/mouse) was administered daily intraperitoneally (I.P), with the first dose of duvelisib injected 24 hours prior to injection of 145-2C11. Plasma IL-6 levels were determined using a mouse IL-6 ELISA (R&D Systems). Injection of 145-2C11 acutely elevated plasma IL-6 >32 fold relative to non-treated controls (4hrs; Control 34.4 pg/mL versus vs. 145-2C11 1088±99.6 pg/mL). Duvelisib significantly reduced mean plasma IL-6 >54% at 4hrs (Vehicle 1088±99.6 pg/ml vs duvelisib 492±99.6 pg/ml, p≤ 0.01) and >78% at 24hrs (Vehicle 220±101 pg/ml vs. duvelisib 49.3±16.1, p≤ 0.01) (Fig 1c) consistent with the effect of duvelisib in our in vitro CAR-T-induced CRS model described above. These studies demonstrate that duvelisib can inhibit CAR-T induced IL-6 production from iDC while having no inhibitory effect on CAR-T. Experiments are currently in progress to further characterize PI3K-δ,γ inhibition in humanized mouse models of CRS and CAR-T efficacy. Our preclinical data suggest that dual PI3K-δ,γ inhibition with duvelisib may represent an attractive alternative to IL-6 receptor antagonists, such as tocilizumab, for the treatment of CAR-T-associated cytokine release syndrome in the clinic which warrants further clinical evaluation. Figure 1. Dose dependent effect of duvelisib on IL-6 secretion in co-culture of CART19, Ramosluc, and iDC after 48 hours (a) Dose dependent inhibition of CART19 mediated cytotoxicity using duvelisib (b) Duvelisib effectively lowers IL-6 plasma level in an immunocompetent BALB/cJ mouse model of CRS (c). (** p≤ 0.01 and ns p>0.05). Figure 1 Disclosures Cooper: Wugen: Consultancy, Current equity holder in private company, Patents & Royalties. Pachter:Verastem Therapeutics: Current Employment. DiPersio:Magenta Therapeutics: Membership on an entity's Board of Directors or advisory committees.