Implications of survival curve slopes on the interpretation of adjuvant therapy results.

Two mathematical models can approximate the survival curves for malignant diseases. The models identify the segments of the survival curve. Also, the hazard function of the curve and the confidence intervals of the curve could be calculated. First, we studied the survival-after-relapse curve of malignant melanoma. The curve of chemo-immunotherapy showed three segments and that for immunotherapy had two segments. The immunotherapy showed its effect in the early period of treatment. Second, the disease-free survival curves for adjuvant therapies of breast cancer were compared. In the Oncofrance trial, a combination of adriamycine, vincristine, cyclophosphamide (C) and 5-fluoro-uracil (F) (AVCF) was superior to a combination of C, methotrexate and F (CMF) in all the periods of the therapy. In Lacour's trial, poly A-poly U was more effective than the no treatment in the middle and late period. In Bonadonna's trial, CMF was superior to no treatment in the early period. Third, the survival curves for immunotherapy versus non-immunotherapy of stomach cancer were analysed. Comparison of the confidence intervals of each curve clarified that no significant difference could be found between them.