30P Protein signature associated with response and progression free survival in late-stage NSCLC patients treated with anti-PD-1 blockade

Use of immune checkpoint inhibitors in cancer therapy has significantly improved response and survival rates of patients with various cancer types including NSCLC. However, beneficial therapeutic effects are not achieved in all patients and predictive as well as prognostic biomarkers associated with response at a timepoint are often failing when outcomes such as progression free survival (PFS) or overall survival are applied. Yet, the use of novel tools oriented towards precision medicine like high-throughput data independent acquisition mass spectrometry (DIA-MS) could address some of these problems. Proteins from plasma samples were extracted and analyzed by capillary flow liquid chromatography coupled to DIA-MS. Proteins were identified and quantification was done with SpectronautTM (Biognosys). Univariate statistical approaches were used to identify significantly changing proteins based on patient response status and progression free survival. Relationships between proteins identified as significant and common for both outcomes were analyzed further using publicly available bioinformatics tools. 125 plasma samples from late-stage NSCLC patients treated with immunotherapy regimens - 75 baseline and 50 after 8-weeks treatment - were analyzed and more than 850 proteins were quantified. Protein signatures associated with response to anti-PD-1 treatment were combined with signatures associated to PFS. In total, 12 common proteins were identified which exhibited long lasting systemic changes on the proteome level. Most of the identified candidates had prognostic characteristics. However, one candidate - TMEM198 (Transmembrane Protein 198), showed both predictive and prognostic capabilities. TMEM-198 is a potential novel biomarker with known association to SWI/SNF (SWItch/Sucrose Non-Fermentable) chromatin remodeling complex and REST (RE1 silencing transcription factor) which acts as an oncogene or cancer suppressor. High dimensional proteomic data can provide dynamic information which is tightly related to clinical features. Here we presented a way how such data can be used for selecting biomarker candidates across multiple clinical outcomes.