Olga Ivanova1, Michael Pratta1, Ivan Azarov2, Kirill Zhudenkov2, Yuri Kozinski2, Xuejun Chen1, Jennifer Sheng1
1Incyte Research Institute, 2Modeling and Simulation Decisions FZ-LLC
Introduction: Itacitinib (INCB039110), a novel, selective oral Janus kinase 1 (JAK1) inhibitor, was evaluated as monotherapy in a phase 2 study (NCT04071366) [1] for the prevention of cytokine release syndrome (CRS) and immune effector cell (IEC)–associated neurotoxicity syndrome in response to chimeric antigen receptor T-cell (CAR-T) infusion. Systemic cytokines, including tumour necrosis factor alpha (TNFa), have been proposed as surrogate pharmacodynamic (PD) biomarkers for the inflammatory response correlating with CRS. Population pharmacokinetic (PK) and PD modelling was applied to evaluate the inhibitive treatment effect of itacitinib on TNFa levels following CAR-T infusion. Objectives: To evaluate itacitinib anti-inflammatory action on systemic levels of TNFa following CAR-T therapy and to differentiate once-daily (qd) and twice-daily (bid) dosing regimens, using a refined mechanistic PK-PD model. Methods: The evaluable phase 2 study PK-PD population included patients receiving CAR-T (axicabtagene ciloleucel) therapy for haematologic malignancies, incorporating part 1 (N=20, open-label; itacitinib 200 mg qd) and part 2 (randomised, double-blind; N=23 itacitinib 200 mg bid vs placebo, N=20 (CAR-T only)). Itacitinib was administered from day -3 to day -1 (pre-IEC) and continued post infusion from day 0 to day 26. The observation time was up to 90 days [1,3]. A stepwise PK-PD model framework was employed for data analysis using a previously reported approach, integrating CAR-T kinetics, PopPK of itacitinib and TNFa PD readouts [2]. High variability in individual kinetics data for CAR-T and TNFa led to poor model convergence, as such the final analysis was based on aggregated PK and PD data. Model refinement was applied to elucidate the best fit from the indirect response models tested to represent the effect of CAR-T and itacitinib on TNFa. Model evaluation included Akaike information criterion, relative standard error of model parameters <50%, goodness-of-fit plots, VPC plots. Model parameters were estimated against the placebo and itacitinib bid cohorts based on similar assay conditions used for CAR-T enumeration. Itacitinib PK was represented by typical population predictions of daily exposure from a previously developed 2-compartment model [7]. An indirect response model with stimulation of TNFa production following CAR-T administration with a physiologically fixed rate of 1.82 h-1 based on previous studies [4,8] was found to best describe the PD data [2]. Ctrough , AUC at D5 and D7, and itacitinib daily area under the curve (AUCdaily) were tested as the exposure metrics for modeling TNFa inhibition. Results: The developed PK-PD model adequately represented CAR-T and TNFa data incorporating the effect of itacitinib. CAR-T dynamics were unaffected by itacitinib administration [6]. CAR-T expansion, contraction, and persistence phases were reasonably described by the kinetic model [5]. Model parameters were identifiable (RSE<27%) The AUCdaily of itacitinib provided the best model performance and robust algorithm convergence. Due to a high variability in TNF levels, a cumulative AUCTNF was selected as a representative marker of inflammatory response, to evaluate the magnitude of treatment effect vs CART-only treatment. Infusion of CAR-Ts resulted in an increase in systemic TNFa levels, and this response was inhibited by the presence of itacitinib. Final analysis was carried out using the AUC between the TNFa profiles for the placebo and active treatment arms (bid or qd administration of itacitinib) as a metric of PD response. The simulated bid regimen suppressed TNFa more effectively than qd, based on AUCTNF metrics: 2905 pg·h/mL (bid) vs 4391 pg·h/mL (qd). The decrease in TNFa concentrations was more pronounced with the bid dosing regimen compared with qd, at baseline (day 0): 2.91 pg/mL (bid) vs 4.56 pg/mL (qd), and at peak concentrations (day 5): 4.56 pg/mL (bid) vs 7.3 pg/mL (qd). Conclusions: The itacitinib treatment effect on TNFa stimulation following CAR-T administration was characterized using a refined integrated PK-PD model. CAR-T stimulatory dynamics were described using a linear drug effect, coupled with a non-linear inhibitory effect by the daily exposure of itacitinib. The itacitinib 200 mg bid dosing regimen was associated with superior TNFa suppression to levels not exceeding baseline (placebo) during the observation time, outperforming the qd dosing regimen.
[1] Huarte E, et al. Clin Cancer Res. 2020;26:6299-6309. [2] Ivanova O, et al. Blood. 2024;144(suppl 1):7129. [3] Frigault M, et al. Blood. 2023;142(suppl 1):356. [4] Larsson J, et al. Eur J Pharm Sci. 2021;165:105937. [5] Stein A, et al. CPT Pharmacometrics Syst Pharmacol. 2019; 8:285-295. [6] Pratta M, et al. Transplant Cell Ther. 2024;30: S181-S182. [7] Ivanova O, et al. ACoP 2024. Abstract W-048. [8] Held F. J Pharmacokinet Pharmacodyn. 2019 Feb 18;46(3):223–240
Reference: PAGE 33 (2025) Abstr 11477 [www.page-meeting.org/?abstract=11477]
Poster: Drug/Disease Modelling - Oncology