LAUSR

6004Background: To date there are no validated predictive tests to inform treatment selection for patients with oropharyngeal cancer (OPC). Currently treatment is decided on disease resectability, clinician preference and patient choice. Methods: Objective To develop a predictive test to select treatment for advanced OPC. Participants Training cohort: 543 cases from 10 cancer centres. External validation cohort: 442 cases from 3 centres. Design Multivariable logistic regression of 8 clinical parameters and 10 biomarkers to develop biomarker-only and composite clinical/biomarker predictive models; subsequently validated on a separate cohort. Biomarkers scored by ≥2 ‘blinded’ pathologists. Outcomes Primary: overall survival (OS). Results: 724 males, 261 females; Median follow-up =8.8 (6.86-10.47) years. More validation cases received surgery (53.5% vs 37.9%, p=0.001) and fewer received chemo/radiotherapy (42%vs67.5%; p=0.001) compared to training cohort. The biomarker-only model performed better than the cl...