Metabolomics predict treatment response to vedolizumab in pediatric IBD: results from the prospective VedoKids cohort study

Abstract Background and aim Metabolites produced through host-microbe interactions may help associate metabolomic signatures with therapy response. We aimed to assess serum metabolites as predictors of vedolizumab remission in children with IBD. Methods VedoKids was a multicenter, prospective cohort study of children initiating vedolizumab at any stage of disease. Serum metabolomic profiling was performed at baseline (W0), week 14 (W14), and week 30 (W30) using quantitative metabolomics based on DI/LC-MS/MS method. Random Forest (RF) models with maximum relevance–minimum redundancy (mRMR) feature selection, integrated clinical and metabolomic features to predict clinical remission at the next time point (ie, week 0 to week 14; week 14 to week 30; week 30 to week 54). Results One hundred thirty-eight patients were included (mean age 13.6 years), 64 with Crohn disease (CD) and 74 with ulcerative colitis (UC). In CD, models achieved an area under the receiver operating characteristic curve (AUROC) of 0.80 (95% CI, 0.68-0.92) for predicting response to vedolizumab at W14, 0.92 (95% CI, 0.85-0.99) at W30 , and 0.83 (95% CI, 0.72-0.94) at W54. Metabolites appeared among the most important features, notably PC ae C36:4, citric acid, and malic acid. In UC, corresponding AUROC were 0.72 (95% CI, 0.59-0.85) at W14, 0.77 (95% CI, 0.64-0.90) at W30, and 0.90 (95% CI, 0.80-1.00) at W54. LysoPC a C26:1, picolinic acid, and serotonin were among the most important metabolites for prediction. Conclusions We identified serum metabolites associated with vedolizumab remission in children with IBD, several of which have been previously linked to IBD pathophysiology.