Ultrasound model may help identify UDCA nonresponders earlier

Ultrasound scans and routine blood tests could help clinicians recognize inadequate treatment response sooner and start additional therapies earlier.

A combination of ultrasound imaging and blood test results may help identify patients with primary biliary cholangitis (PBC) who are unlikely to respond adequately to ursodeoxycholic acid (UDCA) within six months after they start treatment, according to a study published recently in Academic Radiology.

The model distinguished patients with insufficient responses more accurately than approaches based on clinical indicators or imaging features alone. The results also strongly matched with treatment response assessments given at 12 months.

“This model facilitates early identification of high-risk patients, enabling them to benefit from timely treatment adjustments,” the study’s authors said. 

UDCA is the standard first-line therapy for PBC. For many patients it can improve the disease’s prognosis, but up to 40% of patients do not respond to it and need to try a different medication. Currently, whether or not UDCA is working for a patient is assessed after they’ve received the treatment for 12 months.

SM newsletter

Sign up here to get the latest news, perspectives, and information about PBC sent directly to your inbox. Registration is free and only takes a minute.

To help identify these patients sooner, researchers investigated whether two-dimensional ultrasound, combined with clinical measurements, could identify these patients earlier on.

The study included 136 patients, of whom 75 achieved a complete biochemical response to UDCA at six months and 61 had an insufficient response. Researchers divided participants into a training group of 95 patients and a test group of 41. 

Read more about PBC prognosis

They extracted 1,561 imaging features from ultrasound scans and compared six machine learning algorithms. A support vector machine algorithm performed best among the imaging models, achieving an average score of 0.696 during cross-validation and a score of 0.833 in the test group on a scale in which 1.0 represents perfect discrimination and 0.5 represents performance equivalent to chance.

Blood tests identified alkaline phosphatase and platelet count as independent predictors of insufficient treatment response: Higher alkaline phosphatase levels and lower platelet counts were associated with an increased likelihood of an inadequate response to UDCA. 

Combining these measurements with ultrasound features produced the strongest results, with a score of 0.865 in the test group, compared with 0.778 for the clinical model and 0.833 for the imaging model alone. These results suggest the combined model was better at distinguishing patients likely to respond adequately to UDCA from those at risk of an insufficient response.

Among 105 patients with 12-month follow-up data, 95 had the same response classification at six and 12 months, demonstrating strong agreement. The combined model also showed moderate predictive performance under five additional treatment response criteria, with scores ranging from 0.716 to 0.829.

For patients, the findings suggest that ultrasound scans and routine blood tests could eventually help clinicians recognize inadequate treatment response sooner and consider additional therapies earlier. However, the study involved a relatively small patient population, and the model requires testing in larger, independent groups before it can be recommended for routine care.