Imaging of Gastrointestinal Stromal Tumours: The Next Wave of Radiology

At first glance, the article by Barat et al might appear to be a classic review on the use of CT and MRI imaging for the evaluation of gastrointestinal stromal tumours (GISTs). ¹ We might expect the article to review the CT and MRI appearances of these unique tumours focusing on tumour size, location, and imaging characteristics. These findings would typically include enhancement patterns, the presence of calcification, as well as luminal location, and would be discussed as a way of distinguishing these tumours from other tumours including lymphoma, adenocarcinoma, and metastases. Yet, this is but just the beginning of the information provided with imaging. Barat et al provide the next steps in what information can be extracted from the datasets and how this information can be used not only for lesion detection and identification but for decision making in management and predicting outcomes. The article focuses on how we can combine detailed imaging analysis with information from genetic analysis. Eighty to 85% of GISTs harbor activating mutations in the KIT protooncogene and 5-10% harbor mutations in platelet-derived growth factor receptor alpha (PDGFRA) genes, with the 2 mutations being mutually exclusive. Wild-type GISTs represent a small proportion of GISTs with no KIT or PDGFRA mutations. ² These genetic alterations are clinically important since they have a major impact on tumour response to therapy, especially with imatinib. The use of radiomics, which has proven valuable across various imaging studies ranging from pancreatic cancer ³ to lung cancer to breast cancer, is in great part the connection between the imaging studies and the genetic profile and guides how we can use the genetic profile as part of our image analysis.

Barat et al also focus on data acquisition and how dual-energy CT may play a role in monitoring response to therapy and determining when response to therapy is not successful before traditional modes of monitoring. ¹ Dual-energy CT has improved the capability of material decomposition compared to conventional CT, which can potentially enhance our ability to detect subtle histologic changes during treatment. The authors also highlight the potential of cinematic rendering with texture mapping as a post-processing tool for analysis of GIST tumours. Cinematic rendering can enhance the subtle texture changes between GIST and adjacent normal gastric mucosa to improve lesion detection. The pseudo-endoluminal views created from cinematic rendering simulate an endoscopic view that can also assist in lesion detection. In addition, cinematic rendering effectively displays the relationship between tumour and the surrounding anatomy, and can assist in operative planning. ⁴

Barat et al also address, in detail, the potential advances in risk stratification, survival prediction, and response to therapy that can be achieved by detailed analysis of the imaging data with radiomics, texture analysis, and additional post-processing tools now available to radiologists. As we speak about artificial intelligence it becomes clear that support from these systems that can expeditiously analyze large volumes of information, including recent advances in technologies like ChatGPT, shows how quickly our world is changing. ⁵ It is these changes to which we must adapt and rapidly master. Barat et al make it clear that the radiologist must be “fluent” in all aspects of information that can be gleaned from the patients’ imaging data.

Radiology is changing and the days of reading a scan and describing the imaging findings alone are ending. The walls between radiology and the work in pathology, oncology, and molecular imaging are coming down. The radiologist of the future will need more than the ability to detect an abnormality but to understand the information embedded in the images and use this data to help define management decisions as never before possible. The key for radiology and radiologists is how to master these new large domains of information while continuing to excel in what we have always done well. Barat et al make it clear in this article that we are not talking about the changes of tomorrow but the changes that have already begun to arrive and to which we must adapt. In this era of increasing volumes and an overly busy workload it will be challenging for us to meet these changes but in the end, we have no choice. Our patients are depending on us to improve their care and outcomes. We congratulate Barat et al for providing a starting point for the call to rethinking our role in patient care. We are excited about the opportunities.