Two algorithms could better predict response to checkpoint inhibitors

While checkpoint inhibitors have revolutionized immunotherapy, many patients don't respond to the inhibitors, and there are currently no accurate diagnostics to predict who will respond. Two papers in Nature Medicine put forth algorithms that use transcriptomic data to better predict melanoma patient response to checkpoint inhibitors.
In the first paper, researchers from the University of Maryland and NIH's National Cancer Institute developed a scoring system called IMPRES, which predicts response based on tumor expression of immune checkpoint genes.

The authors built the predictor by analyzing transcriptomic data of 108 neuroblastoma patients based on the hypothesis that neuroblastoma spontaneous regression could predict an immune response in melanoma. The researchers identified 15 genomic relationships comprising a comparison in expression levels of two immune checkpoint genes that separated patients with spontaneous regression, indicating no tumor progression, from those with high-risk progressing disease...