Prediction of survival in patients with head and neck cancer

Background. In patients with head and neck squamous cell carcinoma (HNSCC) the estimated prognosis is usually based on the TNM classification. The rel ative weight of the three contributing parameters is often not completely c lear. Moreover, the impact of other important clinical variables such as ag e, gender, prior malignancies, etc is very difficult to substantiate in dai ly clinical practice, The Cox-regression model allows us to estimate the ef fect of different variables simultaneously. The purpose of this study was t o design a model for application in new HNSCC patients. In our historical d ata-base of patients with HNSCC, patient, treatment, and follow-up data are stored by trained oncological data managers. With these hospital-based dat a, we developed a statistical model for risk assessment and prediction of o verall survival. This model serves in clinical decision making and appropri ate counseling of patients with HNSCC. Patients and Methods. All patients with HNSCC of the oral cavity, the phary nx, and the larynx diagnosed in our hospital between 1981 and 1998 were inc luded. In these 1396 patients, the prognostic value of site of the primary tumor, age at diagnosis, gender, T-, N-, and M-stage, and prior malignancie s were studied univariately by Kaplan-Meier curves and the log-rank test. T he Cox-regression model was used to investigate the effect of these variabl es simultaneously on overall survival and to develop a prediction model for individual patients. Results, In the univariate analyses, all variables except gender contribute d significantly to overall survival. Their contribution remained significan t in the multivariate Cox model. Based on the relative risks and the baseli ne survival curve, the expected survival for a new HNSCC patient can be cal culated. Conclusions. It is possible to predict survival probabilities In a new pati ent with HNSCC based on historical results from a dataset analyzed with the Cox-regression model. The model is supplied with hospital-based data. Our model can be extended by other prognostic factors such as cc-morbidity, his tological data, molecular biology markers, etc. The results of the Cox-regr ession may be used in patient counseling, clinical decision making, and qua lity maintenance. (C) 2001 John Wiley & Sons, Inc.