Skip to main content
Premium Trial:

Request an Annual Quote

Molecular Classifier Developed to Detect Cases of Pediatric Irritable Bowel Syndrome


NEW YORK (360Dx) – A team of researchers led by Texas Children's Hospital and Baylor School of Medicine in Houston, Texas has built a molecular classifier to distinguish irritable bowel syndrome (IBS) from healthy controls.

By examining patients' metagenomic and metabolomic data, the team hopes to improve diagnosis and nutritional intervention for patients struggling with IBS, irritable bowel disorders (IBD), and other gastrointestinal diseases.

Clinicians typically recommend altered diets, such as low Fermentable Oligosaccharides, Disaccharides, Monosaccharides and Polyols (FODMAP), or targeted medication to patients struggling with chronic gastrointestinal problems. However, doctors often diagnose irritable bowel disorders based on subjective phenotypic behavior, often leading to misdiagnosis and improper treatment.

"But now since we have a fairly clear handle of what is IBS and not IBS, we can apply microbiome science to add another tool to the chest to diagnose and manage patients," Baylor College of Medicine pathology and immunology professor James Versalovic said.

In a study published last week in the Journal of Molecular Diagnostics, Versalovic, a senior author, and his team examined the relationships between pediatric IBS and abdominal pain with intestinal microbes and fecal metabolites by using a clinical characterization and multiomics strategy. 

As part of the Human Microbiome Project's initial phase, Versalovic's team sought to fill in the information gap regarding the link between gastrointestinal disease and IBS. Believing that the link could be affected by the human gut microbiome, Versalovic's team chose to study functional gastrointestinal disorders, exploring IBS as its primary candidate. 

In the study, Versalovic and his team initially selected a group of 23 preadolescent children identified as having Rome III IBS and 22 healthy preadolescent controls. Participants maintained daily pain and stool diaries for two weeks, followed by clinical phenotype determination. The patients submitted the journal along with stool samples to the lab for genetic analysis.

After extracting stool DNA, the researchers prepared and generated whole-genome sequencing (WGS) libraries using 100 base pair-end libraries and Illumina's HiSeq platform.

Versalovic and his team then prepped the samples for metabolomic profiling and characterization. They removed protein fractions while retaining smaller molecules through several organic and aqueous extractions. The group then further processed the samples using a combination of liquid chromatography, tandem mass spectrometry, and gas chromatography.

"The intent is to take valuable quantities of each patient's sample and use the info to train a computer and develop an algorithm, [which can] be used to classify the individual as an IBS case or not," co-first author and Baylor College of Medicine pathology and immunology professor Numan Oezguen explained. "By using a limited number of measurements, we can possibly diagnose patients based on hard facts that are measured, rather than a biased phenotype examination."

The team then sent the samples to Metabolon for metabolite analysis. The firm ran the samples on its mass spectrometer platform, collecting the mass spectra for each sample. Versalovic's group then compared the spectra with the characteristic spectra of known purified compounds to identify metabolites present in the sample.

Versalovic and his team then evaluated WGS-based taxonomic and functional profiles for different abundances in the samples. They analyzed taxonomic profiles at all levels, from phylum to species, and examined functional data metabolic pathways and enzymatic reactions.

The researchers then developed a subject-species metabolite pathway network by applying metabolites, species, and functional pathway abundance data.

Among the 23 IBS participants, the researchers classified 11 individuals as IBS constipation predominant, 10 as IBS un-subtyped, and two as IBS diarrhea predominant. According to the team, the patients did not differ significantly in age or sex distribution. The team did not find any differences in terms of bowel movement frequencies or mean stool form.

However, the researchers saw that IBS children reported significantly more abdominal pain episodes and greater pain severity than controls, which they noted were highly correlated with each other.

The team produced an average of 116.8 million paired-end reads per WGS family. Taxonomic profiling and differential abundance analysis identified enriched amounts of two bacterial species — Flavonifractor plautii and Lachnospiraceae bacteriaum 7_1_58FFA — in stool communities of IBS children.

Across all subjects, the team identified positive correlations between abdominal pain frequency and severity, and the relative abundances of multiple functional pathways. They also saw positive correlations between pain frequency and severity and the concentrations of steroids and sterols, multiple bile acids, and multiple protein-degradation products.

In addition, the team found 58 metabolic pathways and 182 enzymatic reactions related to amino acid metabolism, phospholipid biosynthesis, and vitamin biosynthesis in the enriched IBS cohort (including carbohydrate metabolism and amino acid metabolism). They saw that high-order metabolite classes also differed in IBS patients, including hyocholate, cholesterol, and thymine.

Selecting the 10 most important differentially abundant metabolites, bacterial species, and pathways, Versalovic's team combined the features into a Random Forest (RF) model and generated a classifier to predict pediatric IBS cases from healthy controls. According to the study authors, the classifier had an area under the curve of .93.

The study authors noted that one limitation of the research is the small population of pediatric patients with specific IBS conditions, which they aim to increase in larger cohorts in further studies. By doing so, they hope to determine the degree that the RF could be used to distinguish IBS from other functional GI tract disorders.  

Oezguen emphasized that while the 10 features his team included in the classifier is not an exhaustive list and other features could be added or replaced, he believes he and his colleagues have developed a robust, novel classifier that is based on biochemical features rather than diary entries or phenotypic elements.

"We don't need the entire [human microbiome] genome for diagnostics, but we need to study it for the discovery phase, to distill and discover key features, and then focus on those key features and develop a tool for the clinic," Versalovic said.

Using the new targeted approach, Versalovic said that researchers in a clinical lab could "reasonably" determine whether a patient has IBS or not within a week.

The researchers believe that they could improve multiomics-based disease classifiers in the future and offer refined diagnostics approaches for coupling patients with GI tract disorders and proper nutritional or medical treatments.

"In our estimation, a disease classifier at or around 80 percent would represent a significant advance in the diagnosis of these functional gastrointestinal tract disorders," the study authors said. "In addition, a classifier may help identify the subpopulations of children with IBS who are more likely to benefit from nutritional interventions."

Versalovic believes the findings will potentially lead to metagenomics-based, data-driven precision diagnostics for IBS and other functional GI disorders. He said the next step is to work with pediatric gastroenterologists to see if they can apply the classifier to improve patient outcome — either through modifying fruit and vegetable consumption or placing individuals on a low FODMAP diet — and hopefully help minimize the impact of IBS on pediatric patients.

Versalovic said that he and his colleagues do not currently hold patents on the strategy and have not entered licensing agreements. He noted that the team's main goal in the study was to pursue a proof of concept. It now aims to develop robust classifiers that could serve as material for IP in the future.  

While in the early stages of developing the classifier, Versalovic said that his team would partner with diagnostic parties to eventually commercialize the classifier into an assay for clinical purposes.

"For now, we are developing a clinical laboratory as part of our microbiome center and as part of our pathology department-based clinical laboratories at Texas Children's," Versalovic said. "We're planning to add a diagnostic test for pediatric patients with GI disorders and collaborate with others for developing tests that are generally applicable in the diagnostic and companion diagnostic space."

By aiming to apply the diagnostic tool with nutritional and future drug intervention, Versalovic hopes that his team can gather insight on chronic abdominal pain, which he said can affect up to 20 percent of the world's population.

Compared to previous studies examining IBS in patients, Versalovic argued that his team is focusing on pediatric IBS, which "is a relatively understudied segment of IBS."

In addition, he noted that "in the context of pediatric IBS, we're developing a new disease classifier that is combined in a way to leverage the human microbiome to improve patient management and diagnosis." 

Versalovic believes that his team is helping to usher in a new era for precision medicine, which he sees incorporates metagenomics-based, data-driven precision diagnostics and therapeutics. With his team's work, Versalovic hopes that researchers one day could apply both microbial and human genetics to produce even more precise work.

"In a general sense, such RF-based disease classifiers may be useful for various GI disorders," Versalovic said. However, "the specific features in this publication are only shown to be predictive of IBS, [and] they will probably not have the same predictive powers for other GI disorders."