Initial clinical findings of a mathematical model to predict survival of head and neck cancer.

Objectives: (1) Identify clinical features that impact survival for head and neck cancer. (2) Determine the individual contribution to mortality of significant clinical features. (3) Develop a web-based calculator to integrate clinical features and predict survival outcome for individual patients.
Study Design: Analysis of a national cancer database. We fit patient data to the binary-biological model of cancer lethality, a mathematical model designed to predict cancer outcome. The model predicts the risk of cancer death, using information on tumor size, nodal status, and other prognostic factors.
Subjects and Methods: Analysis was carried out on a cohort of ~50,000 patients with head and neck cancer from the Survey, Epidemiology and End-Results (SEER) 2009 data set and validated with a cohort of ~1300 patients from an institutional Massachusetts General Hospital/Massachusetts Eye and Ear Infirmary database. We developed a web-based calculator written in JavaScript, PHP, and HTML.
Results: The risk of death due to head and neck cancer increases monotonically with tumor size. Each positive lymph node is associated with ~14% extra risk of death. Anatomical site, age, race, tumor extension, N stage, and extracapsular spread contribute to mortality. The lethal impact of these prognostics factors can be accurately estimated by the Size + Nodes + PrognosticMarkers (SNAP) method.
Conclusions: This predictive cancer model and web-based calculator provide a basis for estimating the risk of death for head and neck cancer patients by assigning values to the lethal contributions of tumor size, number of positive nodes, anatomical site, tumor extension, N stage, extracapsular spread, age at diagnosis, and race.