Deep proteomic analysis of the response to immunotherapy in melanoma patients

Introduction/ Background: Up until recent years patients diagnosed with metastatic melanoma had very poor survival rates. However, recent progresses in the immunotherapy field had dramatically improved their prognosis. Though the success rates are good, most patients do not respond. In order to predict the response to treatment, as well as to improve the success rates by combination therapies, it is important to understand the molecular mechanisms underlying the response.

Materials and methods: Here we set to deeply profile the proteome of clinical samples derived from stage IV melanoma patients that underwent tumor-infiltrated lymphocytes (TIL)-based immunotherapy. Overall we profiled the proteome of 42 patients (20 responders and 22 non-responder) using high-resolution mass spectrometry, and quantified over 9200 proteins with high accuracy. Following that we applied machine learning algorithms and advanced bioinformatic tools to explore mechanisms of response.

Results and discussion: Statistical analysis highlighted the main differences between responders and non-responders, which focused mostly on central metabolic pathways. We further applied machine learning algorithms to obtain a predictive signature for responsiveness, which is comprised of 13 proteins and provides high sensitivity and specificity. To further explore the association between mitochondrial activity and the response to immunotherapy, we studied melanoma cell lines using either metabolic perturbations or CRISPR knockouts of selected metabolic proteins. These perturbations confirmed the observed association between the metabolic changes and the sensitivity to T-cell killing, implying a potential mechanism of response to immunotherapy.

Conclusion: Altogether we characterized the metabolic state that is associated with response and propose a novel association between metabolism and immunogenicity of the tumor cells. Furthermore, this work serves as a basis for further establishment of predictive markers for response.