Staging parameters for cancers of the head and neck: a multi-factorial analysis

Utilizing a multi-variate computer based analysis technique, the 1,400 case head and neck tumor registry of Washington University was examined for possible staging variables as an alternative to the current TNM system. Site, age, characteristics of surgical margins, and absence of neck node metastasis were identified as being predictive of nonrecurrence of cancer. A computer model utilizing the variables was compared to the current TNM system and found to be somewhat better (71.2% vs. 65.2%) in accuracy of prediction of nonrecurrence of cancer. Further analysis suggests that future improvements in predictive ability of the model will have to await the identification of new variables such as those that might arise from the current work in areas of flow cytometry or human stem cell cultures.