II-040 Yannick Hoffert

Model-informed intravenous-to-subcutaneous switching of infliximab in patients with inflammatory bowel diseases

Yannick Hoffert (1), Zhigang Wang (1), Mathurin Fumery (2), Maria Nachury (3), Maëva Bazoge (4), Anthony Buisson (4), Erwin Dreesen (1)

(1) KU Leuven, Department of Pharmaceutical and Pharmacological Sciences, Leuven, Belgium, (2) Unité Peritox, Centre Hospitalier Universitaire de Amiens, Université de Picardie Jules Verne, Amiens, France, (3) INSERM U1286 - INFINITE - Institute for Translational Research in Inflammation, Centre Hospitalier Universitaire de Lille, Université de Lille, Lille, France (4) Université Clermont Auvergne, 3iHP, CHU Clermont-Ferrand, Service d'Hépato-Gastroentérologie, Inserm U1071, M2iSH, USC-INRA 2018, Clermont-Ferrand, France

Objectives: 

Infliximab, the pioneering anti-tumor necrosis factor (TNF) monoclonal antibody, transformed the treatment landscape for inflammatory bowel diseases. Twenty-five years after the intravenously (IV) administered originator drug entered the market, a subcutaneous (SC) formulation of its biosimilar, CT-P13, was added to the therapeutic armamentarium [1].

Therapeutic drug monitoring (TDM) of IV infliximab is increasingly practiced to improve therapeutic outcomes of patients with Crohn’s disease (CD) and ulcerative colitis (UC). Recently, we suggested a TDM strategy to guide the IV-to-SC switch for patients on standard dosing regimens [2].

In the present work, we introduce a software tool to predict infliximab exposure in a wide range of IV and SC dosing regimens. We evaluated the a posteriori predictive performance of the population pharmacokinetics (popPK) model using real-world data.

Methods: 

The infliximab popPK model, developed using pivotal trial data [3], was coded in R (v4.3.0). A software tool was developed using the Shiny package (v1.7.4). Simulations were performed using mrgsolve (v1.0.9). Bayesian forecasting was implemented using mapbayr (v0.10.0), which was used to estimate individual PK parameters through maximum a posteriori estimation.

External validation data were obtained from the prospective, exploratory, multicenter REMSWITCH trial [4]. An infliximab trough concentration before IV-to-SC switch (TCIV,obs) was used to predict an infliximab TC after switch (TCSC,pred). For each patient, the absolute prediction error (aPEi) of TCSC,pred,i was calculated as TCSC,pred,i – TCSC,obs,i. The relative PE (rPEi) was calculated as (aPE/TCSC,obs,i)×100%.

A simulation study was performed to demonstrate how the software tool can guide the IV-to-SC switch. Simulation conditions were

  • Dosing: 10 mg/kg IV every four weeks (q4w) to 240 mg SC q2w
  • Patient: 70 kg, no antibodies towards infliximab (ATI), no coadministration of methotrexate
  • TCIV,obs,i: 17.6 mg/L.

Results: 

A total of 41 patients (21 CD, 20 UC) contributed a TCIV,obs:TCSC,obs pair each (d.d. February 2024; will be expanded to n=133). Median body weight was 66 [range 48–114] kg. Two patients had detectable ATI in the TCIV sample. Two other patients were cotreated with methotrexate. Patients received one of four IV infliximab dosing regimens:

  1. 5 mg/kg q8w (n=16)
  2. 10 mg/kg q8w (n=12)
  3. 10 mg/kg q6w (n=5)
  4. 10 mg/kg q4w (n=8)

followed by one of two SC dosing regimens;

  1. 120 mg q2w (n=40)
  2. 240 mg q2w (n=1)

Median TCIV,obs and TCSC,obs were 6.2 [range 0.2–20.0] mg/L and 11.5 [2.0–20.0] mg/L, respectively. Two patients experienced biological relapse (defined as an increase in fecal calprotectin ≥150 µg/g).

Under the assumption of time constant PK parameters, mean aPE and rPE were +2.0 mg/L [95% confidence interval -7.7–5.0 mg/L] and -10.4% (-95.9%–75.1%], respectively.

In the IV-to-SC simulation case, individual PK parameters (clearance 0.44 L/day, central volume of distribution 2.90 L, peripheral volume of distribution 27.04 L, and intercompartmental clearance 13.01 L/day) resulted in a TCSC,pred of 18.6 mg/L right before the third SC injection. The simulation thus predicts no drop in TC after switching, thus presenting a likely safe switch strategy (considering the general belief that TCs drive efficacy).

Conclusions: 

The in-house developed software tool (available at https://lpmx.shinyapps.io/infliximab/) allows interactive simulations and Bayesian forecasting to predict TC evolutions under various IV-to-SC dosing scenarios. Predictions were unbiased yet precision is likely compromised by variable SC bioavailability or possibly by time-varying PK. Overprediction of TCSC may indicate biological relapse, yet analysis of the full dataset is awaited.

References:
[1] Schreiber et al. Gastroenterology (2021) 160, 2340–2353.
[2] Wang Z et al. Clin Gastroenterol Hepatol (2023) 21, 3188–3190.
[3] European Medicines Agency. EMA/CHMP/548703/2019. Accessed August 17, 2022.
[4] Buisson A et al. Clin Gastroenterol Hepatol (2023) 21, 2338–2346.

Reference: PAGE 32 (2024) Abstr 10813 [www.page-meeting.org/?abstract=10813]

Poster: Drug/Disease Modelling - Other Topics