Monthly Scientific Update – April 2023

Radiomics Models May Help Identify Candidates for Surgery

Most cholangiocarcinoma tumors are unfortunately detected at advanced stages, but surgical resection is sometimes possible for earlier-stage cancers. To better identify patients who may benefit from surgery, researchers are working to develop ways to predict postoperative disease recurrence.

A recent study thus identified variables associated with post-surgical recurrence and aimed to develop a predictive clinical model. The researchers identified four variables independently associated with postoperative recurrence: male sex, microvascular invasion, TNM stage, and CA19-9 levels. However, the use of these parameters in a clinical model did not strongly predict post-surgical disease recurrence.

The authors thus explored an alternative approach called radiomics, which incorporates numerous imaging features in a high-throughput model. Using machine learning, the study identified over 50 relevant radiomic features and selected the ten most significant for further study in multiple predictive models. These radiomics models had more predictive power than the clinical models, with similar results when using both clinical and radiomic variables in combined models.

These results are promising and suggest that radiomic models could be used to help identify cholangiocarcinoma patients who could benefit from surgery. Only a small subset of cholangiocarcinoma patients is eligible for surgery, however, so further research on early detection is essential. New treatments for more advanced tumors are also urgently needed and could make curative surgery a possibility for more cholangiocarcinoma patients.

Kelly Butler

Kelly Butler is an NIH Postbac Research Fellow and the Founding Director of SAFE