Epidermoid carcinoma of the oral cavity and oropharynx: validity of the current AJCC staging system and new statistical tools for the prediction of subclinical neck disease.

Abstract

The 1983 and 1988 AJCC T- and N-staging systems were compared using the case records of 531 patients with primary epidermoid malignancies of the oral cavity. All patients had a minimum followup of 5 years. There were 390 patients with early stage (T1, T2) disease and 141 with advanced stage (T3, T4) lesions according to both the 1983 and 1988 T-definitions: 342 patients manifested no clinical nodes (NO), 189 had clinically evident nodes (N1-N3), and none had metastatic disease. Cox regression analysis demonstrated that the 1983/1988 T-stage definitions differentiated survival successfully (p < 0.001). The 1988 staging system for nodal disease showed a highly significant separation of N2 and N3 when compared with the 1983 system (p < 0.001). Of the 342 patients who were staged N0, 154 had primary neck dissection. Logistic regression predicted the incidence of subclinical disease according to the site and the T-stage of the primary tumor with a sensitivity of 78% and a specificity of 95%. We conclude that the 1988 N-stage definition is a better prognosticator of survival than the 1983 definition. Furthermore, a logistic regression model can be used to predict the probability of subclinical disease in primary oral cavity cancers.