背景:肠易激综合征(IBS)对饮食的反应存在很大差异,差异反应的决定因素知之甚少。
目的:我们的目的是研究可发酵的寡头,di-,单糖,以及IBS患者的多元醇(FODMAP)和麸质。
方法:数据来自一项交叉研究,对FODMAP进行为期一周的干预,面筋或安慰剂。该研究还包括快速激发试验。分子数据包括粪便微生物群,短链脂肪酸,和非靶向血浆代谢组学。用IBS严重程度评分系统评价IBS症状。根据分子和基线问卷数据对IBS症状进行建模,使用随机森林(RF;回归和聚类),平行因素分析(PARAFAC),和单变量方法。
结果:回归和分类RF模型总体预测能力低(Q2≤0.22,分类率<0.73)。在864个聚类模型中,只有2个与集群有显著关联(0.69结论:尽管使用不同的数据分析方法进行了广泛的探索,但无法从临床和分子数据中解释IBS对FODMAP或麸质暴露的差异反应。该试验已在www上注册。
结果:govasNCT0365368931/08/2018。
BACKGROUND: There is large variation in response to diet in irritable bowel syndrome (IBS) and determinants for differential response are poorly understood.
OBJECTIVE: Our aim was to investigate differential clinical and molecular responses to provocation with fermentable oligo-, di-, monosaccharides, and polyols (
FODMAPs) and gluten in individuals with IBS.
METHODS: Data were used from a crossover study with week-long interventions with either
FODMAPs, gluten or placebo. The study also included a rapid provocation test. Molecular data consisted of fecal microbiota, short chain fatty acids, and untargeted plasma metabolomics. IBS symptoms were evaluated with the IBS severity scoring system. IBS symptoms were modelled against molecular and baseline questionnaire data, using Random Forest (RF; regression and clustering), Parallel Factor Analysis (PARAFAC), and univariate methods.
RESULTS: Regression and classification RF models were in general of low predictive power (Q2 ≤ 0.22, classification rate < 0.73). Out of 864 clustering models, only 2 had significant associations to clusters (0.69 < CR < 0.73, p < 0.05), but with no associations to baseline clinical measures. Similarly, PARAFAC revealed no clear association between metabolome data and IBS symptoms.
CONCLUSIONS: Differential IBS responses to
FODMAPs or gluten exposures could not be explained from clinical and molecular data despite extensive exploration with different data analytical approaches. The trial is registered at www.
RESULTS: gov as NCT03653689 31/08/2018.