Design of personalized drug combination treatments for head and neck cancer patients

Oral squamous cell carcinoma (OSCC) exhibit high levels of inter-tumor heterogeneity. Its treatment include surgery, radiotherapy(RT), chemotherapy, PD1/PDL1 inhibitors and EGFR inhibitors (such as Cetuximab, a monoclonal antibody). However, the majority of patients do not respond to the current treatment and the response rate to a single agent (e.g. Cetuximab) is consistently lower than 15%.

We examine OSCC tumor heterogeneity utilizing information-theoretic approach, surprisal analysis. The approach we suggest provides a comprehensive structure of the altered signaling signature in each and every tumor. These signaling signatures may consist of several altered subnetworks, called unbalanced processes. Unbalanced process are a group of proteins that deviate from the steady state in a similar manner due to a certain constraint and thus display correlative aberrations in expression levels. We hypothesize that deciphering the complete set of unbalanced processes in every patient will enable the prediction and optimization of EGFR based treatment. We suggest that by targeting at least one key protein from every unbalanced process the entire altered signaling network in the tumor will collapse.

We analyzed a large dataset including tissues from different types of head and neck cancers. Two OSCC cell lines were used for further experimental validation. We show that the predicted by the analysis drug combinations are highly selective. They are efficient in the inhibition of the predicted cell line and significantly less efficient for another cell line and vice versa. In summary the presented strategy allows to accurately design and optimize personalized EGFR based treatments in OSCC.