Patterns in Cytokine Alterations in Inflammatory Bowel Disease Patients are Suggestive of Underlying Disease Activity

Loading...
Thumbnail Image
Date
2024-04-23
Authors
Potdar, Chinmay
Keyword
cytokine , CCL3 , MIP-1α , IBD , machine learning , multiplex immunoassay
Abstract
Background: Personalized treatment of inflammatory bowel disease (IBD) is a difficult clinical challenge. Large variation in clinical presentation between IBD patients is a major reason for this problem. It is important to investigate various cytokines and immune cell populations as they are known to be the driving force behind the inflammatory process in IBD and may be responsible for the large variation in clinical presentation. The identification of multiple cytokine concentrations as biomarkers for IBD could enable more precise classification, shifting from reliance on phenotypic deductions or limited biomarkers towards empirical evidence. Hypothesis: Patterns in cytokine alterations in IBD patients are suggestive of underlying disease activity. Methods: Both IBD and control patients were enrolled in this study (HSREB 6033229). Patient systemic cytokine profiles were investigated using a 17-plex bead-based immunoassay (n = 23). Clinical data was collected by chart review. Localized tissue expression of a target cytokine identified from analysis of the 17-plex cytokine bead-based immunoassay was investigated using immunohistochemistry (IHC) (n=7). Results: Cytokine immunoassay performed on serum from control (n=8), active IBD (n=5), and IBD in remission (n=10) patients, demonstrated that control patients had a higher mean serum levels of macrophage inflammatory protein 1-alpha (MIP-1α) compared to both IBD active and remission patients. A uniform manifold approximation and projection model based on cytokine concentrations across patients showed differences in clustering between the groups. An extreme gradient boosting model identified MIP-1α as a negative predictor of IBD. A receiver-operator characteristic (ROC) curve generated for MIP-1α as a negative predictor of IBD had an area under the curve (AUC) value of 0.808. Analysis of MIP-1α IHC staining showed distinct differences in the percent of MIP-1α positive cells in the lamina propria of the patients across IBD subtypes as well as across active and remission categories. Summary: This study found for the first time that MIP-1α may be a novel biomarker of inflammation in IBD by combining clinical data with immunophenotyping and analysis utilizing machine learning models.
External DOI