Author:
Mehta Arnav,Rucevic Marijana,Sprecher Emmett,Parikh Milan,Chen Jiajia,Frederick Dennie T.,Woods Elliot,Bi Lynn,Lieb David,Hultin-Rosenberg Lina,Guess Jamey,Park Ryan J.,Schneider Alexis,Michaud William,Miao Benchun,Kasumova Gyulnara,Kim Michelle S.,Bai Xue,Jenkins Russell W.,Klempner Samuel J.,Gonye Anna L. K.,Yizhak Keren,Sade-Feldman Moshe,Liu David,Sullivan Ryan J.,Flaherty Keith T.,Hacohen Nir,Boland Genevieve M.
Abstract
AbstractMost patients treated with immune checkpoint blockade (ICB) do not have durable treatment responses. Therefore, there is a critical need to identify early non-invasive biomarkers of response. We performed plasma proteomic analysis (>700 proteins) at three timepoints on 174 metastatic melanoma patients treated with ICB. We leverage independent training and testing cohorts to build a predictor of immunotherapy response that outperforms several tissue-based approaches. We found 217 differentially expressed proteins between ICB responders (R) and non-responders (NR), including a co-regulated module of proteins enriched in certain NR patients. By analyzing single-cell RNA-sequencing data of tumor biopsies from 32 patients, we dissected the relative contribution of cells in the tumor to proteins in circulation. The majority of proteins in the co-regulated NR module derived from tumor and myeloid cells. Amongst myeloid cells, we identified a subset of tumor-associated macrophages (TAMs) with a suppressive phenotype that expressed high levels of the co-regulated NR module, thus suggesting they are key drivers of non-response signatures. Together, our data demonstrates the utility of plasma proteomics in biomarker discovery and in understanding the biology of host response to tumors.
Publisher
Cold Spring Harbor Laboratory