Synergistic effects of vedolizumab and JAK 1,2,3 inhibitors in Crohn's disease: insights from a systems biology and artificial intelligence-based approach

Abstract

Background and aims: While current therapies for Crohn's disease (CD), including vedolizumab (VDZ) and Janus kinase inhibitors (JAKi), have shown efficacy individually through different mechanisms of action, their combination may offer a more integrated approach, potentially improving outcomes for refractory or severe CD patients. By leveraging in silico systems biology and artificial intelligence, we provide a valuable method to understand the molecular mechanisms underlying this combined therapy.

Methods: CD, VDZ, and JAKi were characterised at protein level by compiling data from extensive literature review and official documents. CD protein effectors were identified and used to construct mathematical models using Therapeutic Performance Mapping System (TPMS) technology. Sampling-based methods were employed to assess the impact of the combined VDZ plus JAKi therapy on CD and to unveil underlying molecular mechanisms.

Results: Disease characterisation identified 4 pathophysiological processes (referred to as motives, M) underlying the manifestation of the disease: intestinal barrier disruption (M1), altered innate immune response (M2), chronic inflammation and predominant Th1/Th17 adaptive immune response (M3), and tissue remodelling (M4). The combination of VDZ and JAKi modulated more effectors of CD (54.8%) than each drug did individually, thereby enhancing the effects of each drug. The combination therapy modulated 45 effectors (36%) through convergent mechanisms, primarily impacting M3, and also provided complementary mechanisms across all motives, but mainly in M1 and M4.

Conclusions: Our results provide a mechanistic rationale for combining VDZ and JAKi as a potential therapy for CD, advancing the development of more effective and personalised treatment strategies.

Keywords: Crohn’s disease; JAK 1,2,3 inhibitors; advanced combination therapy; artificial intelligence; systems biology; vedolizumab.

Figure 1

Schematic representation of the workflow and key steps in the project. The figure represents the Therapeutic Performance Mapping Systems (TPMS) approach to analyse the mechanisms of action (MoA) of vedolizumab (VDZ) plus JAK inhibitors (JAKi) in Crohn’s disease (CD). TPMS is based on systems biology-based models and encompasses three steps (1): molecular characterisations of the disease and drugs through a comprehensive bibliographical revision, from which CD interactome can be constructed using the protein-protein interaction (PPI) human network (2); the learning process of the PPI human network, based on training and validation with known information stored in the training set; this learning is performed with machine learning techniques to construct accurate mathematical models that simulate the behaviour of human physiology; and (3) analysis of specific MoA models for the impact of drug treatment on CD [Jorba 2020; Segú-Vergés 2023].

Figure 2

VDZ, JAKi, and VDZ plus JAKi effects versus CD and each pathophysiological motive. (A) Signal values (fSignal) of all response proteins (CD effectors) after stimulating the model. **p-value < 0.01. (B) Percentage of altered effectors reversed by each drug and the combined therapy in CD and each specific motive. Numbers in parenthesis represent number of protein effectors. (C) Representation of the percentage of protein effectors modulated by each single therapy and by the drug combination. Yellow bars represent vedolizumab treatment alone, orange bars represent JAKi treatment alone and brown bars represent the combination therapy.

Figure 3

Predicted representation of the mechanism of action (MoA) of VDZ plus JAKi in CD. Figure created to represent MoA prediction for VDZ plus JAKi in CD using Graphviz software. All links have been manually reviewed: the numbers of the links correspond to the reference in Supplementary Table S6. Green arrows indicate activation; red lines indicate inhibition; blue lines indicate complex or dual relationships; rhombuses indicate drug targets; broken-lines indicate a node that contains more than one protein, all acting in the MoA together; full-filled circles indicate convergent effectors.

Figure 4

Schematic representation of key pathophysiological mechanisms modulated by VDZ, JAKi or its combination on CD. Motive M1: Intestinal barrier disruption. This figure was drawn using images from Servier Medical Art, which is licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

Figure 5

Schematic representation of key pathophysiological mechanisms modulated by VDZ, JAKi or its combination on CD. Motive M3: Chronic inflammation and predominant Th1/Th17 adaptative immune response. This figure was drawn using images from Servier Medical Art, which is licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).

Figure 6

Schematic representation of key pathophysiological mechanisms modulated by VDZ, JAKi or its combination on CD. Motive M4: Tissue remodelling. This figure was drawn using images from Servier Medical Art, which is licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).