Human Papillomavirus (HPV) contributes to over 70% of all Oropharyngeal Squamous cell carcinomas (OPSCC). In addition to classical treatment strategies (radio-chemotherapy or trans-oral surgery), immune checkpoint inhibitors (ICI) have been utilised to treat patients with recurrent metastases. However, only ~20% of patients benefit from the ICI approach, questioning how and why patients responded differently to ICI-therapy. We hypothesised that interactions between the tumour and its microenvironment inform the relative success of ICI therapies. To test our hypothesis, we developed a cutting-edge spatial-multiomics (spatial transcriptomics (ST) and spatial proteomics; CODEX) approach on tumour tissue derived from an ICI treated HPV+OPSCC patient.
ST was performed on tumour and healthy matched OPSCC samples from a failed Nivolumab (PD-1) therapy patient. This patient received subsequent Pembrolizumab (PD-1) and Lenvatinib (VEGFR inhibitor) therapy but after initially displaying tumour regression, Lenvatinib’s dose reduced coincided with OPSCC resurgence. Unbiased clustering identified 11 distinct clusters based on gene expression profile similarity. These clusters were observed to closely recapitulate the morphological images whereby two data-defined clusters overlayed the annotated tumour site. Differential gene expression (DEG) analysis confirmed these clusters as OPSCC and additionally identified significant proportion of spots in both S and G2M cell-cycle phase, representative of a proliferative tumour. CODEX confirmed the identities of these clusters (Ki67+PANCK+ cells). Significantly up-regulated genes (p<0.001) included various oncogenes (e.g. MYC) and druggable targets including VEGFA. Importantly unlike VEGFA, PD-1/PD-L1 was not expressed in the tumour, supporting our hypothesis that Lenvatinib was likely driving the OPSCC regression. ST was also performed on the recurrent OPSCC, with results displaying shared tumour clustering and similar DEG profiles to the initial tumour.
Our findings suggest spatial-multiomics data can accurately identify the patient’s disease progression and additionally generate a list of clinically relevant drug candidates, allowing oncologists to implement personalised treatment methods to improve patient outcomes.