People with ulcerative colitis (UC), a chronic inflammatory bowel disease, face a significantly higher risk of developing colorectal cancer compared to the general population. The presence of low-grade dysplasia (LGD), which are abnormal or precancerous lesions, serves as an early warning sign. However, only a small fraction of UC-LGD cases progress to cancer, complicating the decision-making process for both clinicians and patients regarding surveillance and potential preventative surgery.
In a groundbreaking study, researchers at the University of California San Diego have demonstrated that artificial intelligence (AI), when combined with biostatistical risk models, can accurately predict which UC-LGD patients are most likely to develop cancer. Published in the journal Clinical Gastroenterology and Hepatology, the study promises to enhance patient counseling, decision-making, and timely follow-up care.
AI and Biostatistics: A New Frontier in Cancer Prediction
The researchers developed a fully automated AI workflow capable of analyzing past medical records, including colonoscopy and pathology reports, of 55,000 patients within the U.S. Department of Veterans Affairs (VA) healthcare system. This dataset is the largest of its kind in the United States, providing a robust foundation for the study.
“Large language models accurately derived colitis-associated colorectal cancer risk factors – such as how big the low-grade dysplasia lesion is, whether there are multiple lesions, and if the colon is extremely inflamed – from the narrative clinical notes themselves,” said Kit Curtius, PhD, assistant professor of medicine in the Division of Biomedical Informatics at UC San Diego School of Medicine and a member of Moores Cancer Center.
Key Findings and Implications
The AI workflow and statistical risk model predictions successfully categorized patients into five risk groups based on four established factors: dysplasia size, lesion resection completeness and visibility, number of dysplastic sites, and severity of inflammation. The model matched real-world patient outcomes with high accuracy for over a decade following diagnosis.
Notably, the AI model classified nearly half of the patients into the lowest-risk group, accurately predicting that almost 99% would avoid a cancer diagnosis within two years. This finding could potentially extend the surveillance interval for low-risk patients, reducing the frequency of medical visits.
“A lot of people are low risk – they have small dysplastic lesions – and it’s been hard to know what to confidently tell these people until now,” Curtius noted. “With this tool, there may be a potential to increase the surveillance interval so patients who are at this low risk don’t have to come back so often.”
Transforming Clinical Practice
The AI model also highlighted that patients with unresectable visible lesions, which cannot be safely and completely removed through surgery, are at a significantly higher risk than many clinicians previously estimated. This insight is crucial for tailoring patient care and ensuring timely interventions.
The study suggests that AI models can seamlessly integrate into clinical workflows, offering precise, automated risk assessments that guide decision-making for both clinicians and patients. This ranges from scheduling the next colonoscopy to considering surgical options, all while reducing the burden on healthcare teams.
“Currently, the process of advising people about levels of risk is a somewhat subjective thing, and doctors don’t have enough data to back up what they feel,” Curtius explained. “This AI pipeline could read the clinical notes and tell you your risk score, rather than just having a list of risk factors and no real way to turn that into a number during a patient visit.”
Future Directions and Broader Implications
Looking ahead, the next steps involve validating the AI tool in patient populations outside of the VA system and incorporating emerging risk factors and genetic information. This could further refine the predictive capabilities of the model.
“We know that genomics play a big part in driving cancer progression,” Curtius said, highlighting the potential for integrating genetic data into future iterations of the AI tool.
The development of this AI tool represents a significant advancement in the management of ulcerative colitis and its associated cancer risks. By providing more accurate risk assessments, it empowers both patients and healthcare providers to make informed decisions, ultimately improving patient outcomes and optimizing healthcare resources.
