Metabolic predictors of response to immune checkpoint blockade therapy

Abstract

AbstractMetabolism of immune cells in the tumor microenvironment (TME) plays a critical role in cancer patient response to immune checkpoint inhibitors (ICI). Yet, a metabolic characterization of immune cells in the TME of patients treated with ICI is lacking. To bridge this gap we performed a semi-supervised analysis of ∼1700 metabolic genes using single-cell RNA-seq data of >1 million immune cells from ∼230 tumor and blood samples treated with ICI. When clustering cells based on their metabolic gene expression, we found that similar immunological states are found in different metabolic states. Most importantly, we found metabolic states that are significantly associated with patient response. We then built a metabolic predictor based on a dozen gene signature which significantly differentiates between responding and non-responding patients across different cancer types (AUC = 0.8-0.86). Taken together, our results demonstrate the importance of metabolism in predicting patient response to ICI.