Signatures from PET and MRI Predict Breast Cancer Outcomes
Image features can provide information about tumor grade, tumor stage, cancer subtypes, disease recurrence, and recurrence-free survival to guide personalized treatment.
Shutterstock/Image Point Fr
Image features that can be extracted from PET and MR scans may have potential as imaging biomarkers for the prediction of breast cancer recurrence-free survival, new research finds. The study, published Aug. 16 in NPJ Breast Cancer, suggests that predictive models based on these image signatures can be used to guide personalized treatment.
“For breast cancer, breast MRI and PET are the most important imaging modalities to stage the cancer so that appropriate treatment decisions can be made,” said senior study author Youngho Seo of the University of California, San Francisco. “Our study extends this concept of using breast MRI and PET from staging to prediction of disease outcomes.”
In cancer management, multiple imaging modalities such as CT, MRI, PET, and single photon emission computed tomography are often used for tumor detection, staging and characterization. Collectively, the imaging data is rich in information that can be extracted for in-depth analysis -- an approach called radiomics that has recently shown promise as a predictive tool.
This emerging technology holds potential for discovering imaging biomarkers and managing disease-specific personalized treatments. But a lingering question is how to determine the optimal use of radiomic features for disease prognosis. There are few studies of radiomics for the same disease across imaging modalities such as PET and MRI, and the added value of these image features remains largely unknown.
“Radiomics is basically finding image signatures that can be extracted from medical images,” Seo explained. “There is very little literature showing radiomics of both PET and MRI is useful in developing predictive models of breast cancer characterization and outcomes. PET and MRI are indeed the workhorses in staging breast cancer, and without exploring radiomic features of both PET and MRI, radiomics research for breast cancer is incomplete.”
To address this gap in knowledge, Seo and his collaborators extracted 84 features from PET and MR images of 113 breast cancer patients and created three subgroups based on the radiomic features. These subgroups were statistically significantly associated with tumor grade, tumor overall stage, breast cancer subtypes, and disease recurrence status. Moreover, PET and MRI radiomics showed high potential for predicting recurrence-free survival. According to the authors, the findings provide optimism for the eventual construction of an effective predictive model based on both PET and MRI radiomics for improved personalized disease management and treatment planning.
“This paper shows the capability of radiomics to harness the data beyond the picture component of the images,” said Gary Whitman, professor of radiology and radiation oncology at the University of Texas MD Anderson Cancer Center in Houston. “The main weakness is that the data in this paper was based on a small number of cases. A next step would be to replicate these findings with a larger, independent data set.”