Precision Medicine: Using machine learning and classification methods to identify immune evasion signatures in high-risk malignancies

Immunotherapies such as immune checkpoint inhibitors (ICi) have changed the therapeutic landscape for many cancers, but the strong association between immunosuppression/immune evasion and poor SCC outcomes implies that early use of such agents will be essential necessitating robust identification of ‘high-risk’ SCC at diagnosis for stratification to receive adjuvant strategies. New technologies such as NanoString and Ion Torrent allow immune-based transcriptomes to be produced rapidly and reliably from formalin-fixed, paraffin-embedded (FFPE) pathology blocks. Pilot data from this group has shown great success in successfully stratifying cancer transcriptomes from cutaneous SCC using machine learning algorithms (decision tree building and linear discriminant analysis). We propose to develop, optimize and validate this approach in SCC from multiple anatomical sites using machine learning and classification to unravel ‘omics’ data and create precision medicine applicable to NHS delivery.