The Study: This study describes a next-generation framework for testing and predicting immune checkpoint inhibitor (ICI) response. In it, broad ranges of tumor types and response predictors from The Cancer Genome Atlas (TCGA) were analyzed to determine what factors were most strongly associated with ICI response. At the heart of the framework is a “triple axis” of ICI response predictors. Each axis clusters predictive factors according to:  tumor neoantigens (e.g., tumor mutational burden),  the immune microenvironment (e.g., immune cell infiltrate), or  actual checkpoint targets (e.g., PD-L1). The single strongest correlate of ICI response was the abundance of CD8+ T-cells (eCD8T), which was followed closely by tumor mutational burden, then the proportion of high PD-1 mRNA expression. In fact, PD-L1 protein expression alone was pretty poor at predicting response. Putting these three together created a trivariate predictive model with a R-value of 0.9 for ICI response. In other words, finding strong predictors of each of the three important aspects to cancer/immune interaction—tumor neoantigens, checkpoint receptors, and immune microenvironment—is remarkably accurate at predicting overall response across a wide range of cancer types.
TBL: It’s likely we’ll see more complex “triple axis” immunotherapy panels in the future that more accurately predict ICI response. | Lee, JAMA Oncol 2019