A virtual ulcerative colitis patient population to explore combination treatments in in silico clinical trials
Rebekka Fendt (1), Venetia Karamitsou (1), Markus Rehberg (1), Klaus Flechsenhar (2), Britta Wagenhuber (1)
(1) Translational Disease Modeling, Sanofi, Frankfurt am Main, Germany, (2) Research, Sanofi, Frankfurt am Main, Germany
Objectives:
Ulcerative colitis (UC) is a chronic inflammatory bowel disease (IBD). Advanced treatment options with biologics only achieve clinical remission in about 30 % of patients. To break this efficacy ceiling, combination treatment can unleash synergies between drugs and pose a new therapeutic strategy. The VEGA trial showed that combination of golimumab (anti-TNFα) and guselkumab (anti-IL-23) was more effective than either monotherapy [1]. We developed a virtual UC patient population for an IBD quantitative systems pharmacology (QSP) model that recapitulates the outcome of the three study arms of the VEGA trial. The generated virtual patient population is a powerful tool to simulate in-silico clinical trials, support decision making during drug development and guide patients to their individual treatment.
Methods:
A previously developed IBD QSP model simulates disease pathways and connects their pharmacological perturbation to clinical outcomes [2]. A global sensitivity analysis identified key parameters that were varied to generate virtual population candidates. These candidates were screened against published ranges of cytokine levels and cell counts [2,3]. Pharmacokinetic (PK) models for anti-IL-23 (guselkumab) and anti-TNFα treatments (golimumab) were implemented into the IBD QSP model to simulate the applied dosing-regimens of the VEGA trial (NCT03662542). A virtual patient population was built from candidates to match the reported remission rates. Remission was defined as Mayo Endoscopic Score (MES) of 0 or 1 at week 12.
Results:
The generated virtual UC patient population consists of 71 virtual patients. The predicted PK for golimumab, guselkumab, and the combination is close to the median of the study population. The simulated remission rates in the virtual UC patient population matched the observed remission rates at week 12 from the VEGA trial (golimumab: 23% simulated vs 25% observed, guselkumab: 28% simulated vs 29% observed, combination: 49% simulated vs 49% observed). Model validation with previously published unseen test data for week 38 [1] indicated that the golimumab efficacy was very well captured during the maintenance phase (treatment after week 12), whereas the efficacy of guselkumab was slightly overpredicted. The simulated population median of the biomarker fecal calprotectin (FCP) fell into the interquartile range of the observed FCP.
The developed virtual population captures all study arms and allows for an in silico head-to-head comparisons of the different treatments. Interestingly, twelve of the 71 virtual patients responded to both monotherapy treatments. So, the virtual population indicates some additivity from the different mode of actions. Eight virtual patients responding to guselkumab (anti-IL-23) did not respond to golimumab (anti-TNFα). Four virtual patients responding to golimumab did not respond to guselkumab.
Conclusions:
We successfully created a virtual patient population matching the VEGA trial outcome. The IBD QSP model with the virtual UC patient population supports decision making during drug development. In silico clinical trial simulations also account for PK variability by randomly matching individual PK parameters with virtual patients from the population. Applications of the modeling framework include optimization of study designs evaluating different dosing regimens, efficacy predictions for new combination treatment approaches or patient subpopulations (e.g., anti-TNFα and anti-IL-23 inadequate responders), and the assessment of best-in-class potential of new drug candidates.
Acknowledgements:
We acknowledge excellent QSP modelling support by Jaehee Shim, Douglas Chung, and Piet van der Graaf (Certara) who extended the IBD QSP model by relevant disease mechanisms and established the initial virtual ulcerative colitis for a combination treatment with Adalimumab (anti-TNFα) and Mirikizumab (anti-IL-23).
References:
[1] Feagan BG, Sands BE, Sandborn WJ, Germinaro M, Vetter M, Shao J, Sheng S, Johanns J, Panés J; VEGA Study Group. Guselkumab plus golimumab combination therapy versus guselkumab or golimumab monotherapy in patients with ulcerative colitis (VEGA): a randomised, double-blind, controlled, phase 2, proof-of-concept trial. Lancet Gastroenterol Hepatol. 2023 Apr;8(4):307-320. doi: 10.1016/S2468-1253(22)00427-7. Epub 2023 Feb 1. PMID: 36738762.
[2] Shim J, Mandema J, Goebel B, van der Graaf P, Rehberg M, Chung D (2022). Predicting Disease Activity Scores in Inflammatory Bowel Disease (IBD) by Connecting Simulated Tissue Biomarkers of a Virtual Population (VPop) to Clinical Biomarkers and Outcomes Using a Novel Algorithm [Poster Presentation]. American Conference on Pharmacometrics 13, Aurora Colorado, USA.
[3] Rogers KV, Martin SW, Bhattacharya I, Singh RSP, Nayak S. A Dynamic Quantitative Systems Pharmacology Model of Inflammatory Bowel Disease: Part 1 - Model Framework. Clin Transl Sci. 2021 Jan;14(1):239-248. doi: 10.1111/cts.12849. Epub 2020 Aug 21. PMID: 32822108; PMCID: PMC7877855.