P266 Predictive tissue-based biomarkers in treatment of Inflammatory Bowel Disease with biologicals and small molecules: a systematic review

Abstract

The treatment landscape of inflammatory bowel disease (IBD) has changed in the past decade with introduction of new drug classes. For most drugs, only 30% of patients achieve long-term remission. Clinical implementation of predictive biomarkers to support precision medicine is lacking. As the interactions of immune system and microbiome in the colonic mucosa may hold the key to the prediction problem, we performed a systematic review to assess the status of tissue-based biomarkers in the era of biological and small molecule therapies. A literature search was performed in Medline, Embase, the Cochrane Library, Web of science and the ECCO abstract database in July 2023. Papers were considered eligible when reporting on biomarkers, extracted from biopsies obtained during colonoscopy, that had predictive value regarding therapy success of a specific non-TNFα biological or small molecule. Two authors independently assessed eligibility and extracted the data. From 10,141 screened records with 155 papers assessed for eligibility, a total of 43 studies were included. The majority was published in the last 5 years (87%). Data on the clinical history of patients and the interval between biopsy and therapy onset was often lacking, in part while 49% of studies were published as conference abstracts. Heterogeneity in therapy success assessment regarding modality (clinical, endoscopic, and/or histologic) and interval (4-58 weeks) was striking. With 28 studies, vedolizumab is the best investigated drug. Of the 16 studies reporting on gene expression, 7 used data from the GEMINI-I/LTS trials. These studies reported a range from a single predictive gene such as TREM1 and OSM to a classifier of 267 differentially expressed genes. Studies looking at the protein level (n=7) focused mostly on target engagement. In 7 studies, biomarker discovery was performed through histological evaluation, mainly looking at mucosal eosinophil counts. One study looking at the microbiome did not find differences between responders and non-responders of vedolizumab. Our systematic review shows that tissue-based predictive biomarker discovery in IBD is a young and dynamic field of research. This is reflected by the heterogeneity in how and when therapy success is assessed. Moreover, the various methods to analyze gene expression data yield vastly different results when applied to the same dataset. We therefore propose guideline development to support clinical implementation of predictive biomarkers.