Gastrointestinal stromal tumors (GISTs) are rare mesenchymal neoplasms that can appear throughout the gastrointestinal system, although they are most prevalent in the stomach (50%-60% of cases) and small intestine (30%-40% of cases), with the esophagus and colorectum being the most uncommon locations.1

GISTs are formed as a result of genetic alterations in 1 or more genes. A mutation in the KIT gene is responsible for about 80% of cases, while a mutation in the PDGFRA gene is responsible for approximately 10% of cases. The remaining 10% are wild-type due to the lack of mutations in either of these genes. Mutations in the KIT and PDGFRA genes have been linked to both familial and sporadic GISTs. These mutations result in proteins and signaling pathways that are always activated, therefore increasing cell proliferation and survival and leading to tumor growth.2

These tumors may be benign or cancerous, although all GISTs have the potential to become malignant and metastasize. In patients with GISTs, contrast-enhanced computed tomography (CT) is the preferred imaging modality for preoperative diagnosis, staging, postoperative follow-up, and therapy response evaluation.3 Nevertheless, predicting the behavior of GISTs in clinical practice is difficult.


Continue Reading

Radiomics is a technique that uses data-characterization algorithms to extract a vast number of characteristics from medical images. These features, known as radiomic features, have the ability to reveal tumoral patterns and characteristics that the naked eye cannot detect.4

Radiomics analysis is a multistep procedure that begins with image acquisition and progresses through lesion segmentation, feature extraction, feature selection and reduction, predictive model construction, and ultimately validation and clinical interpretation of the results. Image data is provided by radiological modalities such as CT, magnetic resonance imaging (MRI), positron emission tomography (PET)/CT, and even PET/MRI. The extracted raw data volumes are utilized to identify distinct pixel/voxel properties using extraction tools.5

After the images are saved, they must be reduced to the key components known as “volumes of interest”. Segmentation can be done manually by professional radiologists, semi-automatically, or automatically by the software. Many features may be extracted after segmentation, and the relative net change from longitudinal images (delta-radiomics) can be estimated. Because of their vast diversity, feature reductions must be done to minimize redundant information. To speed up the procedure, hundreds of distinctive features must be assessed using selection algorithms. It is critical to examine the selected data after choosing the features that are essential for the task. The clinical and molecular (and occasionally genetic) data must be combined prior to the actual analysis since it has a significant impact on what may be concluded from the analysis.5

Radiomics Applications in GISTs

It has been shown that radiomic features may be useful for predicting patient survival, treatment response, and metastatic potential of tumors.6-8

In a review article published in the World Journal of Gastroenterology, Canella and colleagues wrote that they aimed to describe the new advances of radiomics applied to CT and MRI for the evaluation of gastrointestinal stromal tumors, discuss the potential clinical applications that may impact the management of patients, report on limitations of current radiomics studies, and explore future directions of the field.9

They reviewed 15 radiomics research studies on GISTs that had been published before May 2020. Only 4 of the studies were multicentric, and all were conducted in a retrospective population. The number of GISTs included in the studies ranged from 15 to 440 lesions. Except for 1, all radiomics GIST studies used CT imaging to extract features, while just one used MRI.9

The studies included in this review article demonstrated how radiomics can be used in the differential diagnosis of GISTS. Using arterial and venous phase CT images and a texture analysis approach, Ba-Ssalamah et al differentiated GISTs from gastric adenocarcinomas and lymphomas with a high rate of success. Lu et al reported excellent discrimination of GISTs from duodenal adenocarcinomas and pancreatic ductal adenocarcinomas in the periampullary region using whole lesion histogram analysis on contrast-enhanced CT.10,11

The potential of radiomics models for risk stratification in GISTs and for the preoperative prediction of high-risk GISTs compared to conventional visual evaluation with promising results and excellent diagnostic performance. The texture analysis model achieved diagnostic accuracy similar to that of clinical and subjective imaging features for preoperative risk prediction of GISTs12. The histogram-based texture parameters and qualitative analysis of CT imaging features showed a higher diagnostic performance compared to conventional qualitative evaluation by 2 radiologists in the differential diagnosis of low-risk from high-risk GISTs.13

Limitations of Using Radiomics With GISTs

Canella and colleagues state that radiomics has enormous research potential for the improvement of quantitative tumor evaluation. However, they also mention some limitations that challenge its application in everyday clinical practice. These include the standardization problem of the methodology, differences in imaging acquisition, features extraction, and radiomics software, the lack of standardization in CT and MRI acquisition, the peculiar histopathological characteristics of GISTs, and the rarity of GISTs compared to other neoplasms.

Therefore, they indicate the need for further prospective multicentric studies to validate the optimal diagnostic performance of radiomics models provided by retrospective analysis and for optimization and standardization of radiomics software, imaging acquisition, features extraction, and models analysis.

An Era of New Technologies in Health Care

Today, more than ever, a multidisciplinary approach to problem-solving is required. We live in an era in which seemingly unrelated scientific domains are becoming increasingly integrated, and the boundaries between them are progressively blurred.

New information technologies can be used to solve problems in an interdisciplinary context. In recent years, we have seen cooperation between traditional biological study domains and data science specialties transforming medical imaging, lab tests, genomes, and patient histories into accessible, therapeutically helpful insights. Several studies from various fields, including GISTs, have been published as examples of these collaborations, underlining the potential of tools such as radiomics to enhance clinical decision-making. We may expect research findings that are better structured, taking into account the limitations of previous studies, to better represent this potential and meet unmet needs in this field in the near future.

References

  1. Parab TM, DeRogatis MJ, Boaz AM, et al. Gastrointestinal stromal tumors: a comprehensive review. J Gastrointest Oncol. 2019;10(1):144-154. doi:10.21037/jgo.2018.08.20
  2. Gastrointestinal stromal tumor. MedlinePlus Genetics. Accessed January 9, 2022.
  3. Vernuccio F, Taibbi A, Picone D, et al. Imaging of gastrointestinal stromal tumors: from diagnosis to evaluation of therapeutic response. Anticancer Res. 2016;36(6):2639-2648.
  4. Yip SSF, Liu Y, Parmar C, et al. Associations between radiologist-defined semantic and automatically computed radiomic features in non-small cell lung cancer. Sci Rep. 2017;7(1):3519. doi:10.1038/s41598-017-02425-5
  5. Kumar V, Gu Y, Basu S, et al. Radiomics: the process and the challenges. Magn Reson Imaging. 2012;30(9):1234-1248. doi:10.1016/j.mri.2012.06.010
  6. Aerts HJ, Velazquez ER, Leijenaar RT, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun. 2014;5:4006. doi:10.1038/ncomms5006
  7. Nasief H, Zheng C, Schott D, et al. A machine learning based delta-radiomics process for early prediction of treatment response of pancreatic cancer. NPJ Precis Oncol. 2019;3:25. doi:10.1038/s41698-019-0096-z
  8. Coroller TP, Grossmann P, Hou Y, et al. CT-based radiomic signature predicts distant metastasis in lung adenocarcinoma. Radiother Oncol. 2015;114(3):345-350. doi:10.1016/j.radonc.2015.02.015
  9. Cannella R, La Grutta L, Midiri M, Bartolotta TV. New advances in radiomics of gastrointestinal stromal tumors. World J Gastroenterol. 2020;26(32):4729-4738. doi:10.3748/wjg.v26.i32.4729
  10. Ba-Ssalamah A, Muin D, Schernthaner R, et al. Texture-based classification of different gastric tumors at contrast-enhanced CT. Eur J Radiol. 2013;82(10):e537-e543. doi:10.1016/j.ejrad.2013.06.024
  11. Lu J, Hu D, Tang H, et al. Assessment of tumor heterogeneity: differentiation of periampullary neoplasms based on CT whole-lesion histogram analysis. Eur J Radiol. 2019;115:1-9. doi:10.1016/j.ejrad.2019.03.021
  12. Yan J, Zhao X, Han S, Wang T, Miao F. Evaluation of clinical plus imaging features and multidetector computed tomography texture analysis in preoperative risk grade prediction of small bowel gastrointestinal stromal tumors. J Comput Assist Tomogr. 2018;42(5):714-720. doi:10.1097/RCT.0000000000000756
  13. Choi IY, Yeom SK, Cha J, et al. Feasibility of using computed tomography texture analysis parameters as imaging biomarkers for predicting risk grade of gastrointestinal stromal tumors: comparison with visual inspection. Abdom Radiol (NY). 2019;44(7):2346-2356. doi:10.1007/s00261-019-01995-4