PO-512 Gene expression signature comprising distinct stromal and tumor-intrinsic signals predicts response to combined anti-PD1 and anti-CTLA4 checkpoint inhibition

Introduction Innovations in both targeted and immunotherapy (IT) of metastatic melanoma have led to improved responses in a considerable number of patients. For both types of therapies, resistance remains a formidable challenge. For IT, knowledge on the mechanisms behind resistance is an urgent and unmet need for better patient stratification. Gene expression signatures to predict response in baseline samples could help to better stratify patients. It is expected that both tumour cell intrinsic signals and stromal signals play a role. However, gene expression signatures are typically derived from RNA-sequencing data from patients’ tumours, where the distinction between stromal and tumour signals cannot be made. Here, we take advantage of our melanoma PDX platform [Kemper et al. Cell Reports 2016] to dissect the stromal and tumour cell intrinsic signals and relate these to the response to immune checkpoint blockade. Material and methods RNA sequencing was performed on 95 PDX tumours derived from metastatic melanoma [Kemper et al. Cell Reports 2016]. Gene expression data and clinical information of patients treated with anti-PD1 and anti-CTLA4 were downloaded [Hugo et al. 2016 & Van Allen, 2015]. Additionally, we performed RNA sequencing on samples of patients treated with the combination of anti-CTLA4 and anti-PD1 [Blank et al. submitted]. Results and discussions We found that tumour cell-intrinsic and stromal gene expression signatures differentially predicted response to anti-PD1 and anti-CTLA4 immune checkpoint inhibition. Furthermore, a combination of signatures predicts response in patients treated with the combination of anti-PD1 and anti-CTLA4. Conclusion Our PDX platform provided the possibility to computationally dissect the gene expression signals from tumour cells and stromal tissue. These stromal and tumour intrinsic gene expression signatures differentially predict response to anti-PD1 and anti-CTLA4 in baseline samples which can be used for better patient stratification.