Photo Credit: The best photo for all

Metabolomics and machine learning identified serum biomarkers that predict ulcerative colitis extent, offering insight beyond disease activity levels.

By combining metabolomics and machine learning, researchers have identified biomarkers for different disease extents in patients with ulcerative colitis (UC), according to a study published in the Journal of Crohn’s and Colitis.

“Numerous clinical studies of UC have focused on disease activity, neglecting disease extent,” the researchers wrote. “Recent mechanistic studies of inflammatory bowel disease, particularly genotypic studies of patients with Crohn[‘s] disease at different sites, have made us realize that disease extent is also an intrinsic aspect of UC with important clinical significance.”

The study included 189 patients with UC and 30 healthy controls. Among patients with UC, 37 had inactive UC, 31 had proctitis (Montreal classification, E1), 44 had left-sided colitis (E2), and 77 had extensive colitis (E3).

Constructing Prediction Models

Machine learning identified 220 differential metabolites in serum from patients with UC and healthy controls. Further machine learning analysis identified eight essential metabolites for distinguishing patients with UC from healthy controls.

In patients with UC, 23 metabolites differed between those with proctitis and those with left-sided colitis, six differed between those with proctitis and those with extensive colitis, and six differed between those with left-sided colitis and those with extensive colitis, the study investigators found.

“Based on the above metabolites, we constructed prediction models for the extent of UC and found that all five models performed well,” the researchers wrote.

Parameters Across UC Subtypes

According to the results, patients with extensive colitis had significantly lower levels of tridecanoic acid and significantly higher levels of pelargonic acid compared with other patients. Meanwhile, patients with rectal UC had a significantly lower level of asparaginyl valine than other patients.

“This result suggests that asparaginyl valine is associated with local inflammation in the rectum and can serve as a potential biomarker for predicting E1, while tridecanoic acid and pelargonic acid are associated with extensive colitis and can serve as potential biomarkers for predicting E3.”

Although further research into the relationship between metabolic pathways and disease extent is necessary, the findings provide insight into the metabolic landscape of UC, according to the study investigators.

“This study confirmed that the serum metabolomics of patients with different UC disease extents differed at the same level of disease activity,” they wrote.