Model‐Based Patient Selection and Dosing Strategies for HRAS and PIK3CA Dysregulated HNSCC : A QSP Model for Alpelisib and Tipifarnib Combination

Identifying ideal candidates for cancer therapies is challenging, especially with multiple oncogenic variants involved. In head and neck squamous cell carcinoma, the PI3Kα-AKT-mTOR and HRAS-MAPK pathways are frequently dysregulated. This study demonstrates how a quantitative systems pharmacology (QSP) model can help optimize clinical trial design by guiding patient selection and dosing strategies based on oncogenic genotypes. A QSP model was developed to capture the experimentally observed dynamics of the HRAS and PI3K pathways across five molecularly defined patient cohorts in the KURRENT-HN Phase I/II trial (NCT04997902). The model assessed which genotypes would benefit the most from combination therapy with tipifarnib (farnesyl transferase inhibitor) and alpelisib (PIK3CA inhibitor). Simulation results identified the PIK3CA gain of function (GOF) as the genotype most likely to benefit. Virtual population analysis of PIK3CA GOF with dose escalation to 600 mg b.i.d. tipifarnib and 250 mg q.d. alpelisib suggested that a higher tipifarnib dose could enhance tumor response, potentially due to significant dependency of the PIK3CA-mutant cells on mTORC1 signaling. These simulations were consistent with the clinical data. This key dependency can be targeted by tipifarnib by blocking farnsylation of RHEB, an essential activator of mTORC1. Vpop responders showed that reduced intracellular mTOR activity in simulations increased the likelihood of tumor volume reduction. Global sensitivity analysis identified compensatory feedback, tumor proliferation rate, and PI3K-mTOR crosstalk as key determinants of tumor response. This novel QSP application exemplifies an innovative bottom-up modeling approach to support patient selection and dosing strategies for future clinical studies.