Introduction
The use of immune checkpoint inhibitors (ICIs) in the treatment of recurrent metastatic oropharyngeal squamous cell carcinomas (OPSCC) has had limited success, with only around 30% of patients benefiting from the therapy. In this study, we hypothesized that spatial factors in the tumor may play a crucial role in determining the outcome of cancer therapy.
Material & Methods
We conducted spatial transcriptomics (ST) and in-situ multiprotein detection (Phenocylcer) on tissue samples from a patient with metastatic OPSCC. The patient had initially responded to chemo-radio therapy followed by nivolumab ICI. After new tumor resurged, pembrolizumab + lenvatinib treatment had an initial effect, but new oral tumors re-emerged suggesting drug resistance.
Results & Discussion
Using unbiased clustering based on differentially expressed genes (DEG), we identify 11 clusters (CL), in which CL4 and CL5 were real carcinogenic tissue. These two clusters represented distinct metabolic regions of the tumor with different therapeutic implications. We confirmed these findings with Phenocylcer and developed a new cell deconvolution method called Spatial Proteomics-informed cell Deconvolution (SPiD), which outperformed current methods.
Focusing on the real tumor (CL4-5), although PD-1/PD-L1 expression was absent, we identified 9 over-expressed druggable targets and 9 pre-clinical targets that were patient-specific. To rank each drug’s potential success, we measured co-expression of each target ligand-receptor pair (L/R), and prioritized the two main pathways for drug selection, TF/TFRC and VEGFA/NRP1.
The comparison between the pembrolizumab + Lenvatinib responsive and non-responsive tumors showed shared phenotypes with PD-1/PD-L1low and VEGFAhigh expression, suggesting that treatment failure could be linked to the reduction of Lenvatinib dose rather than drug resistance.
Conclusions
We demonstrate the utility of spatial omics in personalizing drug selection and prioritizing targets based on scientific rationale. The use of L/R interactions and pathway analysis provides a promising approach for improving the success of cancer therapy.