Crohn’s disease is a chronic inflammatory disease of the gastrointestinal (GI) tract. Surgery to remove the affected bowel is often necessary, but relapses are common. There has been no way to predict if or when a relapse may occur. Scientists at the Osaka University Graduate School of Medicine, Osaka, Japan, have developed an artificial intelligence (AI) tool which can record and classify tissue images, and a model that uses the images to accurately predict the postoperative recurrence of the disorder. The AI tool enables more intensive and successful treatment of high-risk patients.
The cause of Crohn’s disease is not known. It can occur at any part of the GI tract, from the mouth to the anus. The most common symptoms are diarrhea, cramping and pain in the abdomen, and weight loss. It can severely and adversely affect relationships, self-image, ability to earn a living, and has other unfavorable consequences.
The 10-year rate of postoperative symptomatic recurrence of Crohn’s disease, is estimated at 40%. Although there are scoring systems to identify and assess the severity of postoperative recurrence, no scoring system had been developed to predict whether Crohn’s disease might recur.
The AI tool also reveals previously unrecognized differences in adipose cells and significant differences in the extent of mast cell infiltration in the subserosa, or outer layer of the intestine, comparing patients with and without disease recurrence.
“Most of the analysis of histopathological images using AI in the past have targeted malignant tumors,” explains lead investigators Dr. Takahiro Matsui and Dr. Eiichi Morii, pathologists at Osaka University, in a statement. “We aimed to obtain clinically useful information for a wider variety of diseases by analyzing histopathology images using AI. We focused on Crohn’s disease, in which postoperative recurrence is a clinical problem.”
Sixty-eight patients with Crohn’s disease who underwent bowel resection between January 2007 and July 2018 were included in the study. They were classified into two groups according to the presence or absence of postoperative disease recurrence within two years after surgery. Each group was sorted into two subgroups, one for training an AI model and the other for validation. When the model was tested with unlabeled images, the results indicated that the deep learning model accurately classified the unlabeled images according to the presence or absence of disease occurrence.
Next, predictive heat maps were generated to identify areas and histological features from which the machine learning model could predict recurrence with a high rate of accuracy. The heatmaps showed that the machine learning model yielded correct predictions in the subserosal adipose tissue layer. The subserosal tissue is a layer of connective tissue just below the serous membrane, which lines the GI tract.
Because the machine learning model achieved accurate predictions from images of subserosal tissue, the investigators hypothesized that subserosal adipose cell morphologies differed between the recurrence and the nonrecurrence groups. Adipose cells in the recurrence group had a significantly smaller cell size, higher flattening, and smaller center to center cell distance values than those in the nonrecurrence group. These features, defined as “adipocyte shrinkage,” are important histological characteristics associated with Crohn’s disease recurrence.
“Our findings enable stratification by prognosis of postoperative Crohn disease patients,” the authors explain. “Many drugs, including biologicals, are used to prevent Crohn disease recurrence, and proper stratification can enable more intensive and successful treatment of high-risk patients.”
The findings appear in The American Journal of Pathology.