RESUMO
INTRODUCTION: Pharmacologic therapies for symptoms of gastroparesis have limited efficacy and it is difficult to predict which patients will respond. In this study, we implemented a machine-learning model to predict the response to prokinetics and/or neuromodulators in patients with gastroparesis-like symptoms. METHODS: Subjects with suspected gastroparesis underwent simultaneous gastric emptying scintigraphy (GES) and wireless motility capsule (WMC) and were followed for 6 months. Subjects were included if they were started on neuromodulators and/or prokinetics. Subjects were considered responders if their Gastroparesis Cardinal Symptom Index (GCSI) at 6 months decreased by ≥1 from baseline. A machine-learning model was trained using lasso regression, ridge regression or random forest. Five-fold cross-validation was used to train the models and the area under the receiver operator characteristic curve (AUC-ROC) was calculated using the test set. RESULTS: Of the 150 patients enrolled, 123 patients received either a prokinetic and/or a neuromodulator. Of the 123, 45 were considered responders and 78 were non-responders. A ridge regression model with the variables: BMI, Infectious prodrome, delayed GES, no diabetes (BIDnD), had the highest AUC-ROC of 0.72. The model performed well for subjects on prokinetics without neuromodulators (AUC-ROC of 0.83) but poorly for those on neuromodulators without prokinetics. A separate model with GET, duodenal MI, no diabetes, and functional dyspepsia performed better (AUC-ROC of 0.75). DISCUSSION: This machine learning model has an acceptable accuracy in predicting those who will respond to neuromodulators and/or prokinetics. If validated, our model provides valuable data in predicting treatment outcomes in patients with gastroparesis-like symptoms.
RESUMO
Objective: Identifying microbial targets in irritable bowel syndrome (IBS) is challenged by dynamic microbiota-metabolite-host interactions. We aimed to assess microbial features associated with short chain fatty acids (SCFA) and determine if features were related to IBS symptoms, subtypes, and endophenotypes. Design: We performed an observational study of stool microbial metagenomes, stool SCFA, and IBS traits (stool form, stool bile acids, and colonic transit) in patients with IBS (IBS with constipation [IBS-C] IBS with diarrhea [IBS-D]) and healthy controls. We analyzed associations of microbiome composition with stool SCFA to identify microbe-SCFA relationships that were shared and distinct across groups. We compared gut microbiome-encoded potential for substrate utilization across groups and within a subset of participants selected by stool characteristics. In IBS-D, we compared stool microbiomes of patients with and without bile acid malabsorption (BAM). Results: Overall stool microbiome composition and abundances of individual taxa differed between groups. Increased abundances of several bacterial species were observed in IBS-D including Dorea sp. CAG:317.. Microbes-SCFA relationships varied across groups after accounting for transit and bile acids. Significant microbe-SCFA were common in IBS-D and several SCFA-producing species were inversely correlated with SCFA. Among participants selected by stool form characteristics, functional profiling demonstrated differential abundances of microbial genes/pathways for SCFA metabolism and degradation of carbohydrates and mucin across groups. SCFA-producing taxa were reduced in IBS-D with BAM. Conclusion: Microbe-SCFA associations differ across IBS subtypes and traits. Altered substrate preferences offer insights into functional microbiome traits and could be used as novel microbial IBS biomarkers. KEY MESSAGES: What is already known on this topic: The intestinal microbiota and its metabolites (e.g., short chain fatty acids [SCFA]) modulate irritable bowel syndrome (IBS) pathophysiology. What this study adds: We studied microbe-SCFA associations across IBS subtypes and endophenotypes to demonstrate (1) the intestinal microbiome plays distinct roles across IBS subtypes, (2) microbial substrate preferences vary between IBS subtypes and influences stool form, and (3) microbe-SCFA patterns may reveal key taxa that underlie shared and distinct microbial mechanisms across the IBS spectrum. How this study might affect research, practice or policy: Findings demonstrate that structural and functional features of the intestinal microbiome may represent unbiased microbial biomarkers for clinical and mechanistic IBS subtypes. Further study of these putative microbial targets as well as their interactions with diet- and host-specific traits should be pursued to develop individualized microbiome-based approached to IBS management.