Radiogenomics: A closer look

The radiologic and pathologic methods of the past century are being replaced by recent knowledge about disease pathology and genetic specialty, moving away from the appearance of disease toward the expression of genes that play an important role in risk stratification, diagnostics, monitoring disease progression and projecting survival.

The recent 2013 meeting of the Radiological Society of North America (RSNA) in Chicago featured a scientific session dedicated to developments in radiogenomics featuring two California-based experts Olivier Gevaert, PhD, assistant professor of medicine and biomedical informatics research at Stanford University, and Michael Kuo, MD, from the University of California, Los Angeles. The panel urged attendees to appreciate and gain a better understanding of how advanced imaging and genetics work together for a more comprehensive picture of disease pathology.

“Clearly the field of diagnostic medicine in general is changing and moving beyond the light microscope,” said Kuo. “With sequencing the human genome, which was first published about 10 years ago, we were able to catalogue all of the different genes for the first time and it gave us a new perspective on how they contribute to disease and how they interact.”

The merging of radiology and pathology has been developing over the past few years, and it has broken substantial ground already. “Almost 75-85 percent of the genome can be predicted by imaging alone,” added Kuo.

Interpreting genes through phenotyping has been especially applicable in the case of oncologic imaging. Non-invasive sequencing via imaging may not yet be possible, but scientists can group similar expression profiles together related to morphology, texture and other characteristics by molecular phenotyping. The sheer heterogeneity of disease can be filtered into subclusters of expression with different groups of genes having different patterns in corresponding visual maps that can inform outcomes in pathology, including overall survival. An example of genetic expression picked up by medical imaging is the overexpression of epidermal growth factor receptor and other physiological tip-offs. So far, research in radiogenomics has mapped the patchwork expression of genetic factors involved in breast, prostate, glioblastoma, liver and lung cancers and many share genes, but clusters of expression begin to create a pattern in imaging that is unique to particular tumors.      

“No two tumors are the same and there are differences in morphology and physiology that are captured by imaging,” said Kuo. “There is tremendous molecular diversity."

Mapping this genetic expression looks more like paint on a canvas than a conventional scan of the body. Swatches of color reveal active signature and dominant genetic expression. In the case of blood cancer, radiologists might look in both annotated qualitative as well as quantitative images for morphologic features, genetic expression, and mutation. “We have a tsunami of molecular data going on,” said Gevaert.

Mathematical texture, characteristics of shape and intensity and other semantic features totaling about 180 possible features of disease per patient are brought to life through digitization. In addition, informatics databases are being developed in order to correlate research data with image data from new acquisitions. All of these features are modeled to produce molecular meta-genes. Predictive maps of meta-gene expression can point toward prognoses such as poor survival. Four such semantic features are related to poor survival, whereas one is correlated with good prognosis. Sharpness of tumors from outside to inside, tumor edge characteristics and tumor axis were especially predictive.

The diversity in cell biology—high expression versus low expression of genes—can be leveraged toward the translation of a pathological context, such as a metabolic disorder, and molecular phenotyping could one day inform drug response. Imaging modalities used to map and extract data about tumors and hundreds of variables can be gleaned from CT, MRI, PET and other advanced imaging modalities. This is one area of research that is likely to take up more limelight in years to come.