All looks grey to me.
The Study: Immuno-assays have serious limitations in predicting who does or doesn’t respond to immunotherapy. As an alternative, we’ve recently learned about the Immunoscore, which used CD3 and CD8 immunohistochemistry and digital image analysis to predict responses. Similarly, this study describes a radiomic approach to predicting tumor immune infiltrate and response to immunotherapy. The technique involves creating a 4 mm thick ring around the tumor’s edge (2 mm outside + 2 mm inside). CT data from this ring is then fed into a free textural analysis software called LIFEx. The genomic data in the training cohort came from the French MOSCATO trial, which in part evaluated CD8 immune infiltrate via CD8B gene expression. After multiple training and validation cohorts, the authors were able to generate a radiomic score that was associated with CD8B gene expression, CD8 infiltrate, and ultimately response to immunotherapy. Don’t look for any easy explanation of the imaging features, though, all 78 of them. We can tell you the imaging phenotype most associated with high radiomic scores consisted of homogeneous tumors with discrete borders. CT imaging seems practical enough, but, after reading the methods and supplement 4 times, we can’t say the same for radiomic immunotherapy interpretations.
Bottom Line: This radiomic score can help predict tumor immune infiltrate and response to immunotherapy with a simple CT scan. All we need now is a diagnostic radiomic immunotherapy fellowship. | Sun, Lancet Oncol 2018