Purpose: The goal of the present analyses is to assess the association between different therapeutic approaches and both the probability of achieving a complete response and the risk of death in patients with stage III-IV, inoperable, squamous cell carcinoma of the head and neck (SCC-HN). Patients and methods. Between August 1983 and December 1990, 273 patients with stage III-IV, previously untreated, unresectable SCC of the oral cavity, pharynx and larynx. were included into two consecutive randomized multi-institutional trials (HN-7 and HN-8 protocols) coordinated by the National Institute for Cancer Research (NICR) of Genoa. The HN-7 protocol compared neo-adjuvant chemotherapy (four cycles of vinblastine, 6 mg/m(2) i.v. followed by bleomycin, 30 IU i.m. six hours later, day 1; methotrexate, 200 mg i.v, day 2; leucovorin, 45 mg orally, day 3) (VBM) followed by standard radiotherapy (70-75 Gy in 7-8 weeks) (55 patients) to alternating chemoradiotherapy based on four cycles of the same chemotherapy alternated with three splits of radiation, 20 Gy each (61 patients). In the HN-8 protocol standard radiotherapy (77 patients) was compared to the same alternating program as the one used in the previous protocol but employing cisplatin, 20 mg/m(2)/day and fluorouracil, 200 mg/m(2)/day, bolus, both given for five consecutive days (CF) instead of VBM (80 patients). A single database was created with the patients oil the two protocols. Age at diagnosis, gender, site of the primary tumor, size of the primary, nodal involvement, performance status and treatment approach were analyzed by the multiple logistic regression model and the Cox regression method. The analyses were repeated including the treating institutions as a covariate (coordinating center versus others). Results: The multiple logistic regression analysis indicates that treatment (alternating more so than others, regardless of the chemotherapy regimen used) (P = 0.0001) is more likely to be associated with complete response. In addition, size of the primary tumor (P = 0.004), nodal involvement (P = 0.02) and performance status (P = 0.009) are prognostic variables affecting the probability of achieving a complete response. The Cox regression analysis indicates that treatment, performance status, size of the primary tumor, nodal involvement and, marginally, site of the primary tumor, are independent prognostic variables affecting the risk of death. When the radiation-alone therapy is adopted as the reference treatment, the relative risk of death is 0.58 (95% confidence interval (CI) 0.40-0.84) for alternating CF and radiation, 0.79 (95% CI 0.53-1.16) for alternating VBM and radiation and 1.30 (95% CI 0.89-1.92) for sequential VBM and radiation. When the treating institution is included in the model, a 34% increased risk of death (P = 0.004) is observed for patients treated outside the coordi nating center. Conclusion. In our series of patients with advanced, unresectable SCC-HN, treatment with cisplatin and fluorouracil alternating with radiation was associated with a more favourable prognosis. The role of the treating institution in the modulation of the treatment outcomes was also relevant.
