Fecal microbiome analysis as a diagnostic test for diverticulitis

Abstract

Disease-specific variations in intestinal microbiome composition have been found for a number of intestinal disorders, but little is known about diverticulitis. The purpose of this study was to compare the fecal microbiota of diverticulitis patients with control subjects from a general gastroenterological practice and to investigate the feasibility of predictive diagnostics based on complex microbiota data. Thirty-one patients with computed tomography (CT)-proven left-sided uncomplicated acute diverticulitis were included and compared with 25 control subjects evaluated for a range of gastrointestinal indications. A high-throughput polymerase chain reaction (PCR)-based profiling technique (IS-pro) was performed on DNA isolates from baseline fecal samples. Differences in bacterial phylum abundance and diversity (Shannon index) of the resulting profiles were assessed by conventional statistics. Dissimilarity in microbiome composition was analyzed with principal coordinate analysis (PCoA) based on cosine distance measures. To develop a prediction model for the diagnosis of diverticulitis, we used cross-validated partial least squares discriminant analysis (PLS-DA). Firmicutes/Bacteroidetes ratios and Proteobacteria load were comparable among patients and controls (p = 0.20). The Shannon index indicated a higher diversity in diverticulitis for Proteobacteria (p < 0.00002) and all phyla combined (p = 0.002). PCoA based on Proteobacteria profiles resulted in visually separate clusters of patients and controls. The diagnostic accuracy of the cross-validated PLS-DA regression model was 84 %. The most discriminative species derived largely from the family Enterobacteriaceae. Diverticulitis patients have a higher diversity of fecal microbiota than controls from a mixed population, with the phylum Proteobacteria defining the difference. The analysis of intestinal microbiota offers a novel way to diagnose diverticulitis.

Robust Microbiota-Based Diagnostics for Inflammatory Bowel Disease

“…RF, NSC, ENET, SVM, and PLS-DA have all been successfully applied in human microbiota classification tasks (22,26), as well as in genomics, a field with similar dimensionality challenges (31)(32)(33)(34)(35). We used the following publicly available packages in the statistical software R: randomForest with default settings for the RF classifier; LiblineaR with type "L1-regularized L2-loss support vector classification" and the cost constant tuned by the heuristics provided by the heuristicC function for SVM; glmnet with ␣ values of 0, 1, and 0.5 for L2-and L1-regularized logistic regression and ENET, respectively; caret for PLS-DA; and pamr for the NSC classifier.…”

Enfermedad diverticular: mitos y realidades

“…Los resultados demostraron que la relación Firmicutes/Bacteroidetes y la carga de proteobacterias eran comparables. Sin embargo, una mayor diversidad de proteobacterias y de todos los phyla fue encontrada en pacientes con diverticulitis 34 . Estudios deberán definir el papel de la microbiota en la ED.…”

New medical strategies for the management of acute diverticulitis

“…Using a polymerase chain reaction-based profiling technique on DNA isolates from fecal samples, Daniels et al [22] recently compared the fecal microbiota of diverticulitis patients with control subjects from a general gastroenterological practice. They found that Firmicutes/Bacteroidetes ratios and Proteobacteria load were comparable among patients and controls (p = 0.20), while a higher diversity in diverticulitis for Proteobacteria (p < 0.00002) and all phyla combined (p = 0.002) were found.…”