Patient characteristics compete with dose as predictors of acute treatment toxicity in early phase clinical trials

Purpose: The purpose of this study was to identify patient characteristics that may be risk factors or markers of susceptibility to adverse treatment effects in cancer Phase I and II clinical trials. Patients and Methods: A total of 459 patients enrolled in 23 therapeutic Phase I and II studies at the Fox Chase Cancer Center were included in the analysis. Patient-specific characteristics, medical and treatment history, closes of experimental agents, and graded toxicities were extracted from case report forms. We developed a novel summary measure, the toxicity index (TI), to better discriminate patients on the basis of their overall toxicity experiences. Mixed model ANOVA was used to model TI on the basis of data from all trials using a specific agent. Generalized estimating equations in the context of binary logistic regression were used to model dose-limiting toxicity. Results: Seventeen pretreatment factors, including performance status, alkaline phosphatase, total bilirubin, serum creatinine, and tobacco use, emerged as significant predictors of toxicity as defined by dose-limiting toxicity or TI. Unexpectedly, dose was not always a predictor of toxicity. Even for values within the normal range, the TI identified serum bilirubin and alkaline phosphatase as predictors of toxicity after treatment with docetaxel and alkaline phosphatase as a predictor for toxicity after treatment with irinotecan. Conclusions: Independent of dose, certain pretreatment characteristics, including measures of organ function that are in the normal range, were found to be predictors of treatment toxicity. Because of its sensitivity to differences in overall toxicity, the TI should prove to be a useful tool for identifying predictors of chemotherapy-related toxicity.