Prognostic factors influencing cisplatin-induced emesis. Definition and validation of a predictive logistic model.

Data obtained from 209 patients who entered different prospective randomized antiemetic trials were analyzed to establish the prognostic value of some variables on the control of cisplatin-induced emesis. The antiemetic regimens evaluated included the following: metoclopramide (M) at 1, 2, and 4 mg/kg total dose and a combination of M (4 mg/kg) with dexamethasone (D) (40 mg). Vomiting lasting more than 1 hour (no protection [NP]) was the endpoint selected for the analysis. A logistic model carried out, first on the 110 subjects receiving the M + D regimen and then on the whole sample, selected sex (P = 0.0001), Eastern Cooperative Oncology Group performance status (PS) (P = 0.006), and age (P = 0.01) among the six factors considered. Since the prognostic value of these variables is mostly related to their interaction, three major risk classes were identified. The corresponding NP rates were 26.3%, 42.4%, and 70.8% for the low-risk, intermediate-risk, and high-risk groups, respectively (P = 0.000002). Except for the lowest M level, the observed NP rates were significantly associated to the risk classes, whereas the antiemetic regimen influenced the antiemetic outcome only in the low-risk class. Regardless of the regimen employed, certain patient characteristics, such as sex, PS, and age, significantly affect cisplatin-related emesis and should be carefully considered in planning further studies.