Tumor regression and other prognosticators in advanced head and neck cancers: a sequel to the RTOG methotrexate study

Abstract

The randomized Radiation Therapy Oncology Group (RTOG) Methotrexate trial in advanced squamous cancers of the head and neck has reported no control or survival benefits when the chemotherapy adjuvant was administered to patients just prior to definitive irradiation. The required data collection and outcome reporting among 146 patients bearing oral cavity primaries and 354 patients with oropharyngeal cancers has allowed a multi-variate approach seeking answers to many unresolved questions. As anticipated, the ability to control these squamous cancers is largely a function of size (T & N stage) with a superior clearance among T3-4 primaries of the oropharynx (66%) contrasted to identically staged oral cavity tumors (48%). Adjusted median survival is more than doubled to 26.6 months or 19.8 months among oral cavity and oropharynx patients respectively, when compared to the 8 month median survival when neither primary nor cervical nodes are controlled. Lymph node deposits also impact upon survival, especially among oropharynx patients where the 17.6 month adjusted median survival among N0 patients declines to 11.0 months when the primaries are associated with N3 nodes. Surprisingly, the ability to control nodel deposits of all sizes (N1, N2, or N3) is superior among oropharynx patients when compared with identical oral cavity metastases (e.g. 71.4% adjusted clearance in N3 oropharyngeal deposits versus 46.1% in N3 nodes secondary to oral cavity primaries). Adjustments for maldistribution of advanced N-stages in association with T-4 primary stage eliminated an apparent T-stage effect upon nodal clearance within both anatomic regions. Finally, the association of T and N-stage upon distant metastases was investigated, with the surprising conclusion that neither initial T nor N-stage exerts any apparent influence on the observed 10-12% occurrence. The interrelationship of these various prognostic variables is explored using the Cox and logistic models.